The Peter Attia Drive - #181 - Robert Gatenby, M.D.： Viewing cancer through an evolutionary lens and why this offers a radically different approach to treatment

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

00:00:15.480 my website, and my weekly newsletter all focus on the goal of translating the science of longevity

00:00:19.800 into something accessible for everyone. Our goal is to provide the best content in health

00:00:24.600 and wellness, full stop. And we've assembled a great team of analysts to make this happen.

00:00:28.880 If you enjoy this podcast, we've created a membership program that brings you far more

00:00:33.280 in-depth content. If you want to take your knowledge of the space to the next level at

00:00:37.320 the end of this episode, I'll explain what those benefits are. Or if you want to learn more now,

00:00:41.720 head over to peteratiyahmd.com forward slash subscribe. Now, without further delay,

00:00:47.740 here's today's episode. Everyone, my guest this week is Dr. Bob Gatenby. Bob is a radiologist

00:00:56.200 whose specialty is treating oncology through an evolutionary lens. Now, I stumbled across

00:01:02.700 his work when doing research for another podcast and got completely transfixed down the rabbit hole

00:01:09.540 of what he does. This is a simply fascinating podcast. And we get into what I believe is a

00:01:15.320 very important way that we need to consider treating cancer. And I hope that this podcast will provide

00:01:21.420 both some hope to people who themselves have cancer, but of equal importance inspire some

00:01:25.840 people who are on the front lines of cancer research in exploring another way to think

00:01:30.740 about treating a disease for which, frankly, we haven't had a lot of great success over the past 50

00:01:35.260 years. Hey, Bob, thanks so much for making time to chat today. I've been looking forward to this one

00:01:44.640 for quite some time. Thanks for having me. I'll tell you, I kind of stumbled across your work

00:01:50.200 in preparing actually for another podcast. So I was doing another podcast some time ago and in the

00:01:56.380 process of it, I actually came across an article in Wired about you, which we'll obviously link to in

00:02:03.060 the show notes. And that whole concept just quickly captivated me such that I kind of made a note to my

00:02:10.700 team and said, Hey, at some point, this is someone I'm going to want to interview. And I'm going to

00:02:14.460 want to talk about this entire approach to cancer. And, and lo and behold, that led us to reach out

00:02:20.060 to you. And that's, that's how it came about. But it's interesting that it really started with

00:02:24.580 this sort of rabbit hole you can easily find yourself on in Google, where you're trying to

00:02:29.280 find out about this treatment for this type of cancer and this treatment. And then all of a sudden

00:02:33.400 you see something and I don't know what led me to read that whole article, but I was very

00:02:37.820 captivated by it. And I guess part of it is my own bias, which is my background is very similar to

00:02:45.380 yours. So tell folks a little bit about your background because your path to medicine was not,

00:02:49.780 I guess, the standard path. No, I guess I, I, when I went to college, I wanted to be a physicist

00:02:55.460 and spent most of that time really thinking that I, I should do that. And then I had some great

00:03:05.480 physicist as mentors and kind of came to the conclusion, I'm not smart enough to do this.

00:03:11.440 And I got to, I got to find something else to do. And of course that was a great humanist,

00:03:16.120 the Vietnam War era and going to medical school was considered something that was really,

00:03:21.700 you know, for humanity and something that I, I decided I would, I would do to my regret,

00:03:27.460 by the way, but, but I did, I did end up doing it.

00:03:30.020 So you were at Princeton as an undergrad in the late sixties?

00:03:33.500 Late sixties, early seventies. Yes.

00:03:36.400 And, you know, Richard Feynman, of course, also did his undergrad at, actually, no, no. Did he do

00:03:41.460 his PhD at Princeton and his undergrad at MIT, right?

00:03:44.340 Yes.

00:03:44.880 So was there still the shadow of Feynman there in the late sixties? Cause he had just won his Nobel

00:03:49.940 prize in about 66, right?

00:03:52.320 Absolutely. And Wheeler was still the department chair and he was, you know, one of the great

00:03:57.020 physicists of his generation. Actually my TA won a field medal.

00:04:02.140 Well, subsequently, you know, I was surrounded by really smart people and it was a very enriching

00:04:08.680 time. I mean, it was just, just very stimulating and, and intellectually exciting to be there at

00:04:13.980 the time.

00:04:14.280 And for people who don't know, the fields medal is like the Nobel prize for mathematics.

00:04:18.760 The mathematics does not have a Nobel prize. The fields medal is a substitute, although it comes

00:04:22.780 with two caveats. One, the recipient has to be, I believe under 40, correct?

00:04:27.240 Yes.

00:04:27.480 And the second caveat is it's only awarded, I believe every four years instead of annually.

00:04:31.640 Yeah. And this was Witten, Dr. Witten, who has been very involved in string theory.

00:04:37.280 So you decide to take this detour into medicine, aside from the altruism and the feeling that you

00:04:43.960 weren't brilliant enough to win a Nobel prize in physics, what else drew you to medicine?

00:04:48.780 Well, you know, that was pretty, pretty much it. You know, it seemed like kind of the right thing

00:04:53.340 to do. It was, uh, again, that era was full of wanting to help the humanity and, and, you know,

00:05:00.140 those kinds of exalted ideas about changing the world. So I guess I entered medical school with

00:05:07.300 those things in mind, assuming that I would link the science there to helping people.

00:05:13.160 But now given that, you know, the space race of the 1960s probably attracted some of the best and

00:05:19.740 brightest people into engineering, which you were already in, did you at some point consider

00:05:25.940 staying in engineering, but not pursuing physics?

00:05:30.700 No. I, and I don't know why I can't quite tell you that somehow medicine just kind of drew me.

00:05:36.760 The idea of it was so compelling that I felt like it was, I don't know, something I wanted to do.

00:05:44.100 Now I will tell you that I hated every minute of medical school. Almost from the beginning,

00:05:48.600 began looking at graduate schools to do something else and ultimately decided to stay in it,

00:05:54.460 but was disappointed by what I found there.

00:05:58.440 Yeah. I mean, I think that's another part of your story that I can kind of relate to,

00:06:01.680 which is not that I hated medical school. I, I actually really enjoyed medical school

00:06:05.260 in part because I think the environment I was in, I went to a med school that was in a nice part of

00:06:09.680 the world. I met really, really wonderful friends and I was infinitely happier than any

00:06:15.220 other point in my life. But I realized very quickly that in biology, you couldn't think your

00:06:22.380 way out of every problem the way you could in math or physics or engineering. And I was,

00:06:28.700 I'll tell you a funny story about how much I got humbled in my first semester of medical school.

00:06:32.480 So as you'll recall from engineering, if you understand things from a first principle

00:06:37.420 standpoint, you don't really have to memorize anything. And so it was always really funny when

00:06:42.400 you had these engineering exams where they would allow you to take a cheat sheet in where you could

00:06:46.740 write every formula under the sun on one piece of paper. And I was such an arrogant little schmuck in

00:06:52.740 college that I took such great pleasure in writing only some stupid, obvious formula like F equals MA,

00:07:00.980 because I knew you had to turn the form in with your exam. And I wanted to demonstrate that all I

00:07:06.620 needed to know was Newton's three laws and I could basically derive anything from them. Like you could

00:07:11.440 come up with Coriolis acceleration if you understood calculus. So I get to medical school and first semester

00:07:18.040 is histology among others. And like an idiot, I really believed I could intuit my way through

00:07:26.920 histology without actually having to memorize everything. And I barely passed the exam. And I was

00:07:36.040 like, Oh my God, it's a, it's a new, there's a new world here. I'm going to just have to commit to

00:07:41.580 memorizing a whole bunch of things. So that sort of foreshadowed some other issues I had with my own medical

00:07:48.680 journey. Did you struggle with some of those things as well? Like what was, yeah.

00:07:52.440 Almost identical experience. The other thing is I had 12 years of Catholic school before I went to

00:07:58.300 college and was very familiar with catechism. This kind of idea, rote memorization. And I thought

00:08:05.200 clinical medicine in particular was catechism-like in the sense that there are scripted questions and

00:08:11.120 answers and you're expected to memorize those. And when asked the scripted question, you are required

00:08:17.320 to make the scripted answer. And it's one of the things that I remember is the scripted question

00:08:23.860 was, why do cancers grow? And the scripted answer was because cancer cells grow faster than normal

00:08:29.980 cells. And that's totally wrong. I mean, that's so wrong at so many levels that it's sort of hard to

00:08:37.320 believe. And yet you can ask a medical student these days and some of them will still say that.

00:08:42.120 And so it was less scientific than I expected and more almost like a trade school in the sense that,

00:08:50.760 you know, you learn from the masters these great, on high, they would tell you that this is the way

00:08:55.740 things are and you would memorize that and you'd, you know, go forth into the world with absolute

00:09:01.780 certainty that you're right. And it's very difficult to get physicians as a group to just stop talking

00:09:10.640 dogma. So just think about this for a minute. Does what you're saying make sense? I think that we

00:09:16.420 were just so familiar. So it just became the way to think, which is to memorize. I memorize more than

00:09:24.100 you do kind of thing. And I bring from my memory these little chunks of facts that that really

00:09:30.660 represents the height of intellectual ability. One of the things I find I run into quite a bit is that

00:09:36.100 if you say, you know, what you've been doing for the last five decades doesn't make any sense.

00:09:41.320 It's hard. Of course, it's hard for anybody to accept that. I mean, I get that just from a human

00:09:46.620 point of view. None of us wants to think that we've been doing something wrong, but it's hard to even

00:09:50.980 get people to think about it. And of course, you, you probably are aware of books like the Doctor's

00:09:55.960 Plague and the, and the ghost map that really talked about, and this was in the mid 19th century,

00:10:01.200 that the medical community is a very, very conservative one. And even moving the medical

00:10:06.900 community toward washing hands or, or using a septic technique did not occur overnight. In fact,

00:10:13.320 it occurred over decades. Didn't even occur within the lifetime of the individual who proposed it. So

00:10:18.160 the individual who first proposed it basically dies in an insane asylum, literally in an insane asylum

00:10:24.080 for having been so rejected and cast out by the medical community. That's how long it took to make

00:10:29.080 that the simple transition of washing hands. It's a, it's a fascinating story.

00:10:33.220 I'll share with you just one last story before we leave this, this point, which is by the time I'm

00:10:37.040 in my residency, so now call it, I'm six years, seven years into all of my medical training,

00:10:43.020 really looking for any excuse to draw in some amount of mathematics to some problem solving.

00:10:49.300 And lo and behold, I meet one of the critical care fellows when I'm doing an ICU rotation

00:10:55.040 and we really hit it off. He was an anesthesiologist, but he had a PhD in math and he was the first

00:11:02.360 person who could really put all of the critical care equations into their truest form of physiologic

00:11:10.660 differential equations. So he wasn't the one that just explained it through dogma, but he could really

00:11:15.160 show you the theory. So he and I would stay up really late on call and, and, you know, forego sleep

00:11:21.640 to instead go really deep. And he would pull out math books and physics books and papers, and we

00:11:27.340 would go over this stuff. And that really inspired me. So a couple of months later, when I was doing

00:11:32.040 another ICU rotation, I became sort of dumbfounded at how naively we would dose antibiotics just based

00:11:40.260 on guessing decay times. And I thought, well, surely there's a better way to do this. And lo and behold,

00:11:46.200 there were equations that could really accurately describe how the plasma concentration of gentamicin

00:11:52.120 would decay. And you needed a few more data points, but you could sample these things. And so sure

00:11:57.740 enough, over the course of a couple of weeks, I built a model that would predict the exact right time

00:12:03.420 to be checking for nadir levels to redose. To make a long story short, when I attempted to implement

00:12:10.280 this, it was met with such resistance that I was actually threatened to be fired by one of the

00:12:17.240 attendings at one point. I was actually going to predict that that would, that's what happened.

00:12:23.980 So when you were in medical school, what drew you to radiology or at the time, did you go into

00:12:29.140 radiology first and then radiation oncology, or how did you make that choice? The radiology was a

00:12:33.580 minimum game puzzle. You know, you, you, you, minimum information puzzle where you, you know,

00:12:38.620 sort of get bits and pieces of things to try to put it together. And I really liked that as an

00:12:43.260 intellectual exercise. I don't know, somehow it appealed to me. And I was a very shy kind of

00:12:49.760 retiring kid. Doing with patients was very difficult for me. It was a lot easier to work, to work with

00:12:57.300 films. It intellectually and psychologically met the kinds of things that I like to do. So it just felt

00:13:05.700 like the right thing. And I'm happy I did it because that, that was really one of the few

00:13:09.340 things medical school that I liked. Which I think is very common for people with our background. I had

00:13:15.000 decided early on that I was going to go into surgery. And one of the drawbacks, I guess, of

00:13:19.880 deciding early on what you're going to do is it makes some of the other rotations you go through less

00:13:25.500 enjoyable. But that wasn't true for radiology because once I got to my month of radiology,

00:13:30.840 I was, I mean, categorically obsessed with this field. Because first of all, all of the residents

00:13:38.440 I met were basically engineers, mathematicians, and physicists. So it heavily selects for those

00:13:44.380 people. And an understanding of math and physics gave you an understanding of how the MRI machine

00:13:51.280 worked in a way that you simply couldn't understand that without the background. And I remember thinking,

00:13:57.080 God, I wish there was a way to do both surgery and radiology.

00:14:01.480 Yeah. Yeah. So yeah, it was, I found it very appealing and, and, and enjoyed the residency and

00:14:08.280 did enjoy people that I began working with in that, in that setting.

00:14:12.060 Where was it in your radiology career, if not sooner, that you began to start to think about cancer in a

00:14:18.840 different way?

00:14:19.720 My first job outside of training was at the Fox Chase Cancer Center in Philadelphia. And I hadn't really

00:14:25.640 thought about cancer very much. And of course, we kind of had these very dogmatic views of cancer

00:14:30.980 and what training I did get. But when you work in a cancer center, I mean, you just want to help.

00:14:37.720 I mean, I mean, it just, it felt like you want to make a contribution because this disease is just so

00:14:42.340 awful. So I decided I would, I would spend more time learning about cancer and I would really just get

00:14:48.960 textbooks and read it. And one of the things I started reading was the, the, the journal Cancer

00:14:53.580 Research, which is the flagship journal of the AACR. And one of the things I found was that I would

00:14:59.560 read an article and say, well, this is really a good article and that's really interesting and

00:15:04.540 important. And then I would read another article and have the same response. But then I would try

00:15:09.380 to think of, well, how do these fit together? How do these relate to one another? And, and I

00:15:14.260 couldn't see that. And, and there was no organizing principles that was involved. I mean, they,

00:15:19.580 the authors themselves were not trying to put these together. They were simply making

00:15:23.920 a sequence of observations. Each were really quite separate. And as you know, in the physics world,

00:15:30.720 these things had happened, the planetary motions, for example, Tiger Brahe and others pretty much had

00:15:36.780 the data, but the data was overwhelming and complicated. Kepler developed a kind of geometric

00:15:41.700 interaction. And then ultimately it was Newton who developed first principles that could put all this

00:15:46.880 together. And, you know, similar with the Balmer lines in the early 20th century, the particle

00:15:51.680 zoo in the mid 20th century, all of those required theoretical mastery of it.

00:15:58.040 You could even make the case with, with Einstein's work around the photoelectric effect. I mean,

00:16:03.460 most people think Einstein won the Nobel prize for relativity, but it was actually for the

00:16:07.360 photoelectric effect. And he was not the first person to make the observation. He was simply the

00:16:12.040 first person to put it all together. And I've always found that to be a very illustrative case

00:16:17.640 of what really, what genius really is. It's that ability to assimilate information and pattern from

00:16:25.460 what others can't see. And that, you know, Einstein and others are the examples.

00:16:29.840 So I decided that maybe where I could contribute is to focus on developing first principles,

00:16:36.040 developing a kind of framework of understanding, but recognizing that it really had to be

00:16:40.760 mathematical. I actually spent about a year relearning the mathematics. I'd forgotten it all

00:16:45.400 by then, then worked on developing, you know, so what are the first principles? And I decided,

00:16:52.480 you know, it'd have to be evolution and ecology that all living systems essentially obey the laws

00:16:57.860 of Darwin and therefore cancer must do the same. And so I sat down and I started writing population

00:17:05.020 dynamic equations, looking at cancer, looking at the interaction of cancer with, with normal cells,

00:17:10.340 and then the cancer cells with each other and competing them and that sort of thing. And so

00:17:14.600 that was how I started it. Why were you so convinced, Bob, that this had to be done mathematically as

00:17:21.360 opposed to theoretically, but without the quote unquote cumbersome mathematics that comes with it?

00:17:27.980 I think that I brought with it an appreciation of non-linearity, that human beings think linearly.

00:17:34.160 And when complex systems have non-linear things like feedback loops, that, so the linear thinking is

00:17:41.500 if you do one, you get two, you get two, you do two, you get four, you get four, you do eight,

00:17:46.080 you know, that we're real good. The human brain is very good at that. But non-linear dynamics were

00:17:50.920 really not good at all. A great example, which is also from Philadelphia, was Benjamin Franklin wanted

00:17:57.920 to see a lunar eclipse one evening, but a nor'easter came in. Now, a nor'easter, if you live in

00:18:03.500 Philadelphia, you're, you're familiar with these. They, winds coming from the northeast, hence the

00:18:08.420 name, you know, these are often violent storms. And so it rolled in and he couldn't see the, the

00:18:13.700 lunar eclipse. Now, Franklin, like all scientists of his day, thought that the wind carried the storm.

00:18:20.060 In effect, if you ask a child, how do you, what, how do you think storms move? They will say,

00:18:24.140 well, it must be the wind blowing it because it makes sense. It's intuitively obvious that that's

00:18:28.720 the case. But when he talked to his brother in Boston about the eclipse, it turned out that the

00:18:34.220 storm arrived in Boston after the eclipse was over. So in fact, the storm was going in the opposite

00:18:41.240 direction of the wind. And he was really the first scientist to recognize that obvious, that, that,

00:18:48.380 that something that's intuitively clear must be happening is also wrong. So I think that

00:18:53.860 in cancers, we see non-linearities all the time. And, and again, the feedback with the

00:18:59.280 evolutionary dynamics of resistance, for example, is a good example of that. And we can't intuitively

00:19:05.260 predict those things. We need to actually understand first principles and the underlying mathematics

00:19:09.960 to capture that piece of it. And so I guess I was very involved in that kind of thing. And to me,

00:19:15.760 it seemed obvious that we needed to do the math because the things that were, were being done in,

00:19:21.760 in cancer treatment were often intuitively obvious, but were clearly not working. So I don't know. I

00:19:27.760 mean, I don't, I mean, I may be putting retrospective analysis on that, but at time, it seemed that it

00:19:34.340 really had to be understood mathematically. And of course, from a physics background, that's just kind

00:19:38.920 of a natural, you know, the theory has to be fundamentally about mathematics.

00:19:43.000 This is just sort of a fun aside question. Why do you think that evolution gave us as humans,

00:19:50.800 the ability to understand linear systems quite well, and absolutely no capability to understand

00:19:57.740 non-linear systems? So for example, it's clear that we don't understand hyperbolic discounting.

00:20:04.100 We just can't do it. Is it simply that evolution wasn't optimizing for that problem? And it really

00:20:10.200 didn't, when it came down to reproduction and survival, linearity was sufficient?

00:20:15.740 That would be my guess, that, that what we need to know to survive and proliferate is sufficiently

00:20:20.320 linear that we can probably, that's probably all that was needed. But I'd have to think more about

00:20:25.900 that. I mean, what's, what's linear in the world that is so important, but, but my sense is that

00:20:31.920 relatively simple things that, that are related to eating and running away from predators and that sort

00:20:37.880 of thing and running after mates, probably sufficiently linear that that was really all that was

00:20:43.480 necessary.

00:20:45.340 So as you're getting deeper into the mathematics of the biology, you're probably struck by the fact

00:20:51.780 that you don't have a lot of colleagues in this space, right? There's not a conference for theoretical

00:20:55.620 biologists, right?

00:20:57.300 No. And similar to the response of your colleagues to your model, pretty much all of my colleagues

00:21:03.980 hated it, thought the whole thing was ridiculous. But were they even qualified to understand it?

00:21:09.140 Probably not, but just on an intuitive level, they just couldn't understand why you would do this.

00:21:15.480 And it's funny because if I had a dollar for every time someone said to me that cancer is too

00:21:20.080 complicated to model, I, you know, I wouldn't have to apply for grants anymore because that was kind

00:21:25.240 of the prevailing wisdom. And I mean, the irony of that is that the argument itself is self-defeating.

00:21:31.100 If it's that complicated, then you have to have mathematics. There's no way that the human brain

00:21:36.900 is going to understand complex systems without sufficient, you know, mathematics. Unfortunately,

00:21:42.960 I think in there was a certain arrogance that you're also taught as physicians, which is that,

00:21:48.940 well, it's too complicated to model, but my superior intellect will be able to sort of master this

00:21:55.200 and figure it out. I mean, I always got the feeling that there is that second clause in that

00:22:00.100 statement that went unsaid, but was part of this kind of idea that this confidence that we can

00:22:07.460 understand this and, you know, step back and we can take care of this without the mathematics.

00:22:12.800 So I think it takes some humility to say, well, I really need to look at the math models and the

00:22:19.160 computer simulations to understand how I think this is going to happen. I don't know that we physicians

00:22:24.660 have been taught humility with sufficiently well to accept that.

00:22:30.480 Another great example of that is look at other incredibly complex systems like economics. And of

00:22:37.900 course, I don't know what the stats would be, but certainly a sufficient number of the Nobel prizes

00:22:43.660 awarded in that field are fundamentally based on mathematics, right? Whether it be game theory or

00:22:48.760 otherwise, obviously a lot of them are behavioral as well. But nevertheless, I don't think anybody is

00:22:53.520 suggesting that the models are sufficient in economics. In other words, that you can take

00:22:59.360 an economic model, you can plug in all of the initial conditions and it will tell you the answer.

00:23:04.300 If unemployment is this, if the rate of home price appreciation is this, if inflation is this,

00:23:10.320 here are 50 starting variables, put them in the model and it will spit out GDP growth 10 years from

00:23:16.460 now. I don't think anybody is so delusional to believe that that's true, but it still doesn't minimize

00:23:21.240 what the model can do for your understanding of the system.

00:23:24.700 We like to talk about hurricane modeling and weather modeling in general, which is a classic

00:23:29.880 example of how to master a complex dynamic system. You can be pretty sure that you can predict what's

00:23:35.380 going to happen in the next 24 hours. After that, the complexity as well as stochastic changes are going

00:23:41.400 to degrade the accuracy. So at each day, you get more data and you just keep predicting forward.

00:23:48.920 It's not necessary, at least again in sort of cancer modeling, to say what's going to happen 10 years

00:23:55.600 from now. It's just you need to know what therapy should I use today and for the next three months,

00:24:01.540 say, and then after that, we'll get more data, we'll start again. But sometimes people expect more of

00:24:08.260 that, of it than that. And like you said, that's this idea that we should be able to predict the

00:24:12.840 entire course is not realistic. You know, I'm glad you brought up hurricanes and weather, because

00:24:17.700 as you note, they're some of the most complicated models out there. And that has to do with the fact

00:24:23.780 that they behave in many ways as these, you know, like Lorenz curbs, right? So they are chaotic systems.

00:24:29.220 And because they are so, so, so sensitive to initial conditions, they don't really behave well

00:24:36.240 outside of very, very narrow Delta T windows. And, you know, as we, I remember still, I remember in

00:24:41.760 college when we began studying chaos, you know, the first example you learn is about the butterfly

00:24:46.920 that flaps its wings in Tokyo that leads to the storm two weeks later in New York. And this is

00:24:52.940 getting way ahead of ourselves because we're going to come back and go through it from an evolution

00:24:55.720 perspective. But just at the outset, before I forget to ask this question, where would you put

00:25:00.860 cancer models on the relative spectrum of here's a really well-behaved model on one side, you know,

00:25:08.860 some linear regression model that works perfectly because you have infinite past data and the future

00:25:15.080 scenarios don't deviate. So that's like a monkey model. And then at the other end of the spectrum,

00:25:19.620 you have the weather predicting hurricane predicting model as the most chaotic. If that's a one in a 10,

00:25:27.160 where does cancer biology behave? I think probably eight or nine. I think it's more on the chaotic.

00:25:33.600 I think there's a lot of stochasticity. There's also a lot of heterogeneity. And I think those

00:25:37.840 things make it difficult. So it's harder to predict, but not infinitely hard. I mean, I think we can,

00:25:45.700 we can do this, but we always have to have a level of humility and understanding that,

00:25:50.860 first of all, evolution is very clever and likes to embarrass you. You can easily crash and burn.

00:25:55.980 But it's not random either. There's predictability to it. And finding that point between them where

00:26:03.360 you can predict with reasonable certainty and also sort of hedge your bets in ways that even if things

00:26:10.340 don't go exactly as you plan, you can still benefit the patient from early recognition that it's not

00:26:16.640 going the way you planned, recalibrating models, rethinking the underlying dynamics, and then going

00:26:22.160 forward. As opposed to, here's your model, just take one a day for the next 10 years and you'll be

00:26:30.080 fine. I don't think we can be at that point. So let's go back to that first year where you've

00:26:37.060 relearned the mathematics. Presumably at some point you say, I want to look in nature and see where

00:26:43.620 mathematics has already come up with an elegant way to describe a similar problem. Is that what led you

00:26:50.580 to the study of predator-prey models? Just the eco-evolutionary dynamics in general?

00:26:57.040 It's very appealing because, of course, there are living systems. The ecology of swamps and that

00:27:02.700 sort of thing is very complicated. And yet they can more or less master these complex relationships.

00:27:10.580 So that was appealing to me. It was also appealing to me because, you know, when you work in a cancer

00:27:16.220 center, the cancer almost takes on a persona, like an evil entity kind of thing, almost magical in its

00:27:24.800 ability to overcome anything that the physicians do. And if you talk to oncologists, many of them will

00:27:30.760 have that. They may not say it, but they have that kind of sense of it. And to me, just saying that

00:27:38.800 cancers have to obey Darwinian laws, that they're not magic, they're not unfathomable.

00:27:44.660 They simply are really good evolutionary machines. And we can master this. We understand evolution.

00:27:53.140 And we can get on top of this. It's not something that we cannot understand. Because before that,

00:27:57.780 it just felt like we have no basis to understand this. It just happens and we do this and this

00:28:03.800 happens. And so this, to me, gave it some deterministic qualities, some cause-effect

00:28:09.160 relationships that, at least to me, were comforting. Like, we could deal with this.

00:28:14.840 So explain to folks how the standard, relatively simple predator-prey models work.

00:28:22.100 The swamp is too complicated because you might have multiple predators and multiple prey. But let's

00:28:26.320 start with first-order differential equations, second-order differential equations, where you've

00:28:31.860 got one, you know, you've got an isolated ecosystem where you've got foxes and hares. How does that

00:28:37.600 dynamic work? Well, you know, the hares convert local vegetation to babies, to baby hares. So they

00:28:45.400 reproduce. The foxes or whatever the predator is going to eat the rabbits. And of course, so then you

00:28:52.460 get these population rises and falls because if the predator eats too many of the prey, then its

00:29:00.340 population will tend to decline. Its population declines, the predator population expands. And so

00:29:06.780 you can see this cycling effect that goes on for prolonged periods of time. And there at least are some

00:29:13.320 data that have suggested that this is the case. But as with everything that's living, it's always more

00:29:19.600 complicated than that. There's always various factors that come into play. But at least that was

00:29:24.980 really, I think, probably the first of really kind of recognized population models that began to be

00:29:33.060 applied to nature. It's interesting to see that. And predator-prey models are things that we use a

00:29:39.160 little bit like for the immune system, where the immune system is kind of a predator and it's chasing

00:29:44.860 after the cancer cells. But of course, there's important differences in that. The predator eats

00:29:52.900 the prey and gains substrate from that. Whereas in the immune system, it kills the cancer cells,

00:30:01.160 but it loses substrate, doesn't eat them up. And in fact, if anybody's eating anything, it's probably

00:30:06.760 the cancer cells are swooping up elements of their brethren that, you know, the macromolecules that

00:30:12.400 are getting spilled into the environment. Again, it's one of those things that it's an appealing

00:30:17.020 model, the predator-prey model in immunotherapy. And yet there are important distinctions that you

00:30:23.060 have to recognize that make the biology different and in some ways can give advantages to the prey that

00:30:28.780 you don't really expect. So when you move on to a more complicated system like the swamp,

00:30:33.560 right, where you've got, you know, you've got everything from algae to bacteria to small fish,

00:30:43.320 and then you have to deal with how much sunlight is coming in and what's the temperature. I mean,

00:30:49.080 now it starts to look a little bit more like the human body, where why is there an algal, you know,

00:30:55.980 bloom here that basically consumes all of the oxygen and rapidly kills all the fish versus

00:31:02.920 a system that can be somewhat in a sustained setting where you never fully get rid of the

00:31:07.840 algae, but the fish can live. And there's kind of a beautiful chain of carbon fixation that goes from

00:31:14.840 algae to fish. How did you get to the point where you could look at that and say, we can now model this

00:31:22.480 for human cancer, given that this is far more likely how it behaves? Again, it's, you know,

00:31:27.940 simplifying. The best we can do is sort of a cartoon. And it's interesting to think about why.

00:31:32.920 I mean, ecologists that are looking at a new ecosystem will begin by asking a very simple

00:31:40.160 question. What's the birth rate and death rate of each species that's present? We don't know that

00:31:45.120 in the cancer. I mean, astonishingly enough, that's not data that we get. What's the carbon cycle? What's

00:31:51.140 the iron cycle? What's the, you know, nitrogen? What are all these cycles? How can we, can we watch

00:31:55.900 these substrate pass through individuals? And how does that work? We don't get that. So,

00:32:01.600 so we call it just kind of turn pale when we, when they start asking these questions

00:32:05.600 and realize that we have, that cancer biologists have never thought in those terms. I mean,

00:32:10.920 something as simple as what is the, what is the birth rate and the death rate of the cancer

00:32:15.320 population? It's astonishing to me that we don't know that, that, that, and things like that, you

00:32:21.760 know, what's the growth rate of the tumor? What are we kind of dealing with it even in a first order

00:32:26.200 kind of estimate? We often don't do. It's kind of astounding to me that we do things very crudely.

00:32:33.960 You know, the evolutionary biologists and ecologists take a far more sophisticated view

00:32:37.920 of these interactions than we do. An interesting fact is that if you were a pesticide manufacturer,

00:32:44.300 you are required by law to submit a resistance management plan. You have to identify what are the

00:32:52.040 mechanisms of resistance? And how do I plan to prevent that from occurring? You can, you can

00:33:00.020 introduce cancer drugs. I mean, well, cancer drugs are routinely approved without any knowledge of

00:33:07.900 what the resistance mechanism is, much less how you're going to manage that in a patient. So again,

00:33:13.200 this, this odd kind of disconnect that in some ways, I think the ecologists, evolutionary biologists

00:33:19.520 have pushed ahead of us and we're just still trying to catch up with them in terms of, of understanding

00:33:25.200 of taking their sophisticated models and applying them to cancer. So let's talk about the ecologic models

00:33:32.000 of pests and pesticides, because that's something that I think gave you a big insight, correct?

00:33:39.240 Yes. One of the things that I had run across was just a kind of a story on the, on the internet

00:33:45.900 news somewhere about the diamondback moth. And the story was that it was first recognized,

00:33:51.980 I think it's in Indiana, somewhere in the Midwest in the mid 19th century. And this, the diamondback

00:33:56.820 moth has, has the, I guess, the honor of having received and struck by every pesticide developed

00:34:03.280 in the modern era. And what it has done is absolutely nothing. In fact, it's spread all

00:34:09.600 over the, the diamondback moth spread all over the country to Europe and to Asia and it's just

00:34:14.480 everywhere. And in the 1980s, they were uncovering diamondback moths that were not susceptible to any

00:34:22.640 known pesticide. I mean, they had become resistant to all of them. To me, this was, it was interesting

00:34:27.740 because, so you begin to sort of look at this whole process and, and of course, farmers for,

00:34:33.600 for centuries use pesticides freely. And the idea is you dump as much of it on your fields as you can.

00:34:40.200 You want to get rid of as many of the pests as you can. But what people realized is that by doing

00:34:45.420 that, you're, you're selecting for resistance. Explain to people why that's true, because it is

00:34:49.980 counterintuitive. I think most people would say, well, gosh, if I'm a farmer and I've got my acre of

00:34:57.700 corn here and there are a million moths that have descended on this acre, I want to get every one

00:35:07.000 of them eradicated. And the best odds of doing that, wouldn't that just be using all of the

00:35:12.560 pesticide I can just shy of killing my own crop? Yes. And that, that's one of those things that is

00:35:19.080 intuitively appealing, but not necessarily true. And the reason is that you're dealing with a very

00:35:25.900 large population. And this is not a uniform population. This is biology, which means that

00:35:32.000 there's, there's heterogeneity within that population. And there are moths in that population

00:35:38.560 that are extremely sensitive to the pesticide. And there's going to be moths in that, that population

00:35:43.220 that are not very sensitive to the pesticide. Okay. So let's put some numbers to it. So I just gave

00:35:48.040 you a million moths, you dump the best pesticide imaginable on them. What if we assume that there's

00:35:55.820 a spectrum and 20% of them die at the first whiff of the pesticide, 60% of them eventually get clubbed

00:36:06.240 to death, but 20% of them, I don't know, they, they seem somewhat immune. Is that, let's just use

00:36:13.720 those numbers, right? Sure. So short-term you did good. Because you got rid of, you got rid of 80%

00:36:19.620 of them. Yeah. Right. Now dump pesticide on them again. So that population that's that 20% now has

00:36:27.560 a whole field that's open to it. It has no competitors. So it can rapidly expand and it's

00:36:34.840 taking its genome with it. And so its offspring will also be somewhat heterogeneous, but they will

00:36:40.340 definitely be shifted more toward the, the resistant ones. So you, now you, you dump your

00:36:45.920 massive amounts of pesticide and maybe you get 5% of them that die. And maybe you get 20% of them

00:36:51.420 that, you know, eventually die, but 80% of them don't. And, and you just keep doing it. So year one

00:36:58.260 or, you know, season one looks good. And, and again, intuitively, you know, like a really good idea,

00:37:06.600 but the long-term effect is that now you've got pests that you can't control anymore. You,

00:37:12.760 you have a pesticide that doesn't work. And so in the longer term, what you've done is you've created

00:37:21.380 a species you can no longer control. Now you have to find a new pesticide and there's always some

00:37:27.460 cross resistance. And in any large biological population, there is sufficient heterogeneity

00:37:34.240 that it's most likely that you're going to have resistance. So if it worked, and part of this is

00:37:42.140 say, well, yeah, okay. But how do you know that there's resistance? And it's simple. If it worked,

00:37:49.140 there wasn't resistance, but you know, from, from experience and over and over and over again,

00:37:54.600 what you found was that you use this large dose pesticides, you can get a short-term gain,

00:38:01.680 but in the long term, it doesn't work. So believe your eyes. So this actually, you know, again,

00:38:10.100 ecologists that were, you know, were way ahead of us here. They said, well, there's got to be

00:38:14.360 something different to this. And so in the Nixon administration, this is a long time ago, same time

00:38:19.100 that they started the Warren Cancer, using large doses of pesticide was essentially not allowed. And

00:38:25.680 they started to do this thing they called integrated pest management. And the idea with this

00:38:30.700 is that you recognize that you can't eradicate the pest, that, that there's simply no, just

00:38:38.320 historically, there's nothing that's ever done that before. So let's assume that's not going to

00:38:43.360 happen this time. Let's give enough pesticide that we're going to knock this population low enough

00:38:51.360 that they're not going to do very much crop damage, but don't do more than that. There's a number of ways

00:38:56.700 you can do that. So for example, they would take a field and they would put pesticide on three quarters

00:39:01.640 of the field and leave one quarter alone. So you would knock this population down, but you would not,

00:39:08.620 to this last quarter of the field, you're applying no selection for resistance. And so these guys would

00:39:15.660 then move out to the rest of the field and, you know, you eventually get this low population,

00:39:20.640 but you're not selected for resistance. There's another bit of this, and that is that

00:39:25.780 the resistance costs something. To be able to deal with a toxic pesticide or toxic drug,

00:39:32.640 you know, you have to have the molecular machinery necessary to deal with it. You have to be able to

00:39:36.860 repair your DNA. If it's damaged, you have to be able to pump it out, which is what a lot of them do.

00:39:43.060 I mean, there's a number of different mechanisms, but all those mechanisms have a cost. It doesn't

00:39:47.460 necessarily have to be a big cost. And in the cost benefit ratio, when you've got a lot of this drug

00:39:54.160 around, those guys are clearly, you know, the benefit greatly outweighs the cost. But when there's

00:40:00.320 no drug present, then the benefit does not outweigh the cost. And so now the guys that are not resistant

00:40:05.940 don't have to carry that machinery around with them, are more fit. And so there's this then subtle

00:40:13.060 competition that goes on. So you'll be selecting for resistance in, let's say, the three quarters of

00:40:17.820 the field. But when the ones that are sensitive move into that area, they have the benefit of not

00:40:23.640 having this pesticide being applied to them anymore. There's no selection for resistance. And their

00:40:29.600 fitness advantage will be such that when the pest population comes back to something close to what

00:40:35.460 it was at the beginning, it's as sensitive as it was in the beginning.

00:40:39.940 Will it actually get better? I mean, is there a scenario here under which, if I'm understanding

00:40:44.880 you correctly, the moths that will be able to best resist the pesticide, that comes at a cost.

00:40:56.540 And the cost is molecular machinery that is only there to allow them to survive the toxicity of the

00:41:05.600 pesticide, but not reproduce better or acquire food better. So the more quickly you can get non-pesticide

00:41:14.260 resistant moths back into their ecosystem, who presumably in the absence of pesticide now have a

00:41:21.240 fitness advantage over them because they don't have the extra baggage they don't need in a non-pesticide

00:41:27.380 world, they'll actually out-compete them. So is there a scenario where in year one,

00:41:33.860 20% of the moths are resistant to the pesticide, but if you manage it correctly in year five, that

00:41:40.960 number is 10% because they have actually been out-competed by the other moths?

00:41:45.900 Yeah. So where were you 20 years ago? Because only recently has your scenario come up. And I guess

00:41:52.520 we're getting a little ahead of ourselves, but we've done a study now in men with prostate cancer,

00:41:57.300 or we've used a drug. And what we did was that we gave the drug until the tumor responded to 50%

00:42:05.680 of its pretreatment value, pulled it away, let the tumor come back up. But again, similar to this tool

00:42:12.680 we talked about, there's no pesticide being applied, there's no drug being applied, there's no selection

00:42:16.200 for resistance. And so the sensitive guys are supposed to come back up and so grow at the expense

00:42:22.860 of the resistance. Now, for a very long time, our models said that what would happen is that every

00:42:31.260 time the sensitive population came down, the resistant population would go up a little bit.

00:42:36.520 And when the sensitive population started going, we took the drug away, let the sensitive population

00:42:41.480 grow, the resistant population would plateau. And what we then assumed is that at each cycle,

00:42:49.400 you would have this plateau. So step-wise, it would just keep going up. And at some point,

00:42:54.960 you would lose control inevitably. All of our models said that if we did this somewhere between

00:43:01.500 two and 20 cycles, the treatment would fail. Wow, that's a huge variance, though, two to 20 cycles,

00:43:07.040 that could be the difference between a few months and a few years. Why such a difference?

00:43:11.600 At the time, we didn't know what the fraction of resistant population was at the beginning. So that was

00:43:17.200 probably the biggest thing. So if it's 1%, that takes longer than if it's, let's say, 20 or 30%.

00:43:22.600 And there are some other things like we weren't really entirely sure what the proliferation rate

00:43:27.040 was. And again, even things like birth and death, we just estimated those. So we did this pilot trial

00:43:34.300 to see if this worked. So we gave this kind of intermittent therapy, and we compared it to patients

00:43:40.060 that just got the standard maximum tolerated dose until progression. The difference between the

00:43:45.900 groups was 16 months. So the median time progression for standard of care was 14 months, which is

00:43:52.200 pretty much what's in the literature. For the adaptive therapy group, it was 30 months, which

00:43:57.300 was great. But then we said, but of the 20, four patients are now out five years, and they're still

00:44:05.240 cycling. And this is patients with metastatic prostate cancer?

00:44:08.260 That's correct. So what the models said, and this now gets very nerdy, but perhaps the most

00:44:16.100 important thing that came out of this trial was the number seven, because the ratio of the fitness

00:44:22.000 measure for the sensitive cells was sevenfold that of the resistant cells. And we had estimated at two

00:44:29.980 or three. So it's a big difference.

00:44:32.460 And it's a big difference in favor of the healthy cell.

00:44:35.700 Exactly. And so what we see then is that when the sensitive cells go down, resistant cells go up,

00:44:43.240 sensitive cells go up, we expected them to plateau.

00:44:46.180 But in fact, they're going down.

00:44:47.520 They go down. And what we found was that if you, in fact, had three cycles where you hit that sweet spot

00:44:54.720 exactly, the resistant cells progressively went toward extinction.

00:45:00.080 So three successive cycles is a critical enough mass that you can drive that resistant cell,

00:45:07.160 presumably to a place where, and by the way, is it limited by substrate? Is that the most

00:45:11.320 important factor that is creating the fitness differential?

00:45:15.140 Yes, we think. Now it's going to be, they're competing for space and substrate. And of course,

00:45:20.840 the tumor environment is one in which usually the substrate delivery is very poor because the,

00:45:27.620 you know, the blood flow isn't very good. So substrate is probably the most limiting factor.

00:45:32.460 But again, we don't know this. I mean, this is, again, one of those things that I wish we knew more

00:45:36.980 about because in some ways we're just working in the dark. But, and so when we went back and looked

00:45:43.540 at these four patients that have gone far beyond anything that you would expect for this treatment,

00:45:49.780 there's very rare that someone would go five years on this treatment using standard of care.

00:45:54.780 What we found was that each of them had these three sequences, cycles, where we, where the model

00:46:01.940 predicted that they drove the, the resistant population to extinction or something very close

00:46:07.980 to extinction. And none of the others had the three cycles?

00:46:11.260 No. And that was the key thing that we learned from this, and which is one of the great advantages

00:46:16.060 of having the math models, because you can self critique. Now, now you don't just take the,

00:46:20.680 the results of the cohort and say, well, this is nice. We did 16 months better. We can also look

00:46:26.440 at what we did and say, what did we do right? And what did we do wrong? And what the model said was

00:46:32.020 that we had built into the, into the protocol, some components that as a result, we did not hit those,

00:46:39.840 those three consecutive cycles perfectly. And universally, we over-treated. If we had cut back on

00:46:47.380 therapy, we would have done still better. And the models predicted that every patient

00:46:52.900 in both cohorts in this study, we could have gained control of the tumor indefinitely.

00:46:58.940 What was the other cohort?

00:47:00.280 The other cohort received standard of care.

00:47:03.340 Oh, oh, oh. But this was not randomized.

00:47:05.340 That's correct. It was a contemporary treatment.

00:47:07.320 And what were you following? Was your metric PSA?

00:47:09.900 PSA. That was probably where the biggest mistake we made is that we, we were following the PSA,

00:47:16.640 but then we required that when, when the PSA went down, that it would be confirmed by radiographic

00:47:21.780 studies. And we were getting the PSA every month, but you're constrained by insurance to get radiographic

00:47:30.020 studies every three months. And so what would happen is that the PSA would get to 50%, but we didn't do,

00:47:36.100 we didn't change therapy until we got the-

00:47:38.520 The radiographic finding.

00:47:40.120 Which meant that-

00:47:40.940 You were lagging.

00:47:41.920 For two months, they kept getting treatment, and they kept driving down. So we were killing off too

00:47:47.240 many of the resistance of the sensitive cells, and therefore, they were not able to-

00:47:51.920 Yep. You were impeding their fitness. Yep.

00:47:54.000 Of course, this is what the model said is, you guys are idiots. You know, if you just stayed with the PSA,

00:47:59.900 you would have gotten far better results than you did, and you got pretty good results anyway.

00:48:03.200 Your question goes to something that we'd honestly never thought about before, which is that we

00:48:11.960 thought we could use the sensitive cells to control the resistant cells.

00:48:17.740 You can actually use the sensitive cells to destroy the resistant cells.

00:48:22.080 To eradicate the-

00:48:23.200 Now, I, you know, never thought about that, but this is now opening up whole new avenues of therapy,

00:48:31.920 at least theoretically, I mean, at least potentially, because there's a lot of cases where

00:48:36.820 the initial therapy for cancers gets great results. I mean, you know, they talk about remission,

00:48:43.060 tumors going into remission, and of course, what happens is the tumor eventually comes back.

00:48:48.140 And what we know then is that the resistant cells are there, but they're there in a very small number,

00:48:53.740 and you can then estimate that they must be way, way, way less fit than the sensitive cells in the absence

00:49:01.860 of treatment. So that fitness difference is, it must be very large. So using a number like seven

00:49:08.440 as the ratio is not unreasonable. In fact, it's highly likely. And so now we've got this situation where

00:49:16.140 maybe we can start to think about using these very effective therapies. Right now, what we do is we

00:49:23.600 give the therapy, tumor kind of disappears, can't see it. And what do we do? Well, we keep giving the

00:49:29.140 same therapy. We give it until progression. Well, as we're doing that, we've knocked the tumor to the

00:49:34.420 ground, but we just keep giving it the same treatment. And of course, what we're-

00:49:39.200 So we're just closing the gap. We're reducing the fitness ratio from seven, six, five, four,

00:49:44.960 three, two, one, and it'll flip on you. Yeah. So now suddenly we've got, all we've got is resistant

00:49:50.720 cells. I can keep treating that all I want. That's not going to work. But these are highly vulnerable

00:49:55.860 to additional perturbations. So the thing is, instead of just giving this treatment, getting the

00:50:03.540 cancer populations to something that's very, very small, and then just keep treating them,

00:50:07.760 at that point, hit them again. Do something different. Apply additional perturbations.

00:50:14.120 One of my colleagues likes to say, this is, we're like boxing here. You know, you knock your opponent

00:50:18.900 down, and you go to your corner. You know, the referee counts, and when the opponent's back up,

00:50:25.820 then you start again. And, you know, this is, it's not a boxing match. It's a knife fight. This is to the

00:50:31.220 death. And that means that when you knock your opponent down, you don't stand back and wait. You go

00:50:36.700 attack. And I think that we've, that's an approach, and we've called that an extinction approach that

00:50:42.000 we think is something that has not been explored yet. But it's interesting that if you look at

00:50:48.140 anthropocene extinctions, if you look at how our species has killed off other species, this is the

00:50:54.280 way it happens. There's usually an initial perturbation. But then after that, when you get a small

00:51:00.000 population, you get a sequence of smaller perturbations that drive it to extinction. It's a

00:51:05.000 two-step kind of process. What's interesting is that you cause extinction through a sequence of

00:51:12.240 perturbations, none of which by itself could cause extinction. In cancer, we're always looking for the

00:51:18.060 magic bullet, the one that will eradicate the cancer and not the normal cells. And for a century, we've

00:51:25.040 been looking for magic bullets, but maybe all we need is a series of pretty good bullets. So that's a

00:51:31.000 different sort of thinking process. But it is hard to sell that, especially to a medical community in

00:51:40.000 which new drug development is what's most incentivized. So using the same drug, in the case

00:51:47.160 of our protocol, it's using a drug called abiraterone, which is off patent. Nobody's making any money on

00:51:53.340 this. So there's no pharma company that's going to support the clinical trials. It's kind of an

00:52:01.180 orphan drug. So there's no natural supporter for this. And the same is true with a lot of these

00:52:07.740 other things that we're really just talking about using drugs that already exist, but using them

00:52:11.760 better. Again, there's just not a natural support for it within the medical community.

00:52:19.480 I want to detour for a second, Bob, and talk about how this logic fits into an area that frankly,

00:52:26.880 I've never considered until now. I've just taken it for granted, which is the use of antibiotics and

00:52:31.120 bacteria. So standard thinking is you have a bacterial infection, you give the antibiotic,

00:52:39.460 and there's some magic number, which is you're going to give it for X number of days. It's a 10-day

00:52:44.500 course. Now, three days into that 10-day course, the patient is defervesced. They feel completely

00:52:50.800 better. The tissue culture is negative, but we continue to give the antibiotic for seven more

00:52:57.580 days because our fear is, or what we were told was, if you don't, you will generate resistance.

00:53:04.400 But based on what you're telling me now, that might not be true. In fact, it might be that giving

00:53:09.840 that seven additional days of antibiotic might be exactly the thing that is leading to resistance.

00:53:16.440 Is that, am I misunderstanding? I think that's certainly true. I mean, I don't work with

00:53:20.660 infections, but there are people that are looking at this exact issue. How do you prevent the

00:53:27.960 development of resistance of bacteria? And worldwide, it's a big problem with malaria and tuberculosis and

00:53:34.680 things like that. And so the World Health Organization actively uses evolution-based

00:53:39.280 mathematical models to plan how they deliver treatment. And a lot of it is, as I understand

00:53:46.660 it, and again, this isn't my area, is that they are trying to kind of do this kind of adaptive approach

00:53:53.420 where you just keep the resistant population low. Now, there are some exceptions where my understanding

00:54:01.060 is some people have advocated regional attempts to eradicate a population. And again, a little bit

00:54:07.460 like the Anthropocene extinctions. So Bob, is one difference, I guess, the more I think about this,

00:54:13.880 between antibiotics and bacteria versus chemotherapy and cancer? I mean, you could live with billions of

00:54:20.520 cancer cells in your body and be totally fine. Is that true with bacteria? Is that true? Could you have

00:54:27.060 a billion MRSA floating around your bloodstream and still be fine? Or do we need complete extinction

00:54:32.240 there in a way that we don't with cancer? You know, I don't know. We have to get someone that's

00:54:38.460 expert in this. The one thing to remember is that by the time you have a cancer, it's defeated your

00:54:44.320 immune system. Whereas the bacteria often are still very different from our normal cells, and therefore,

00:54:51.840 I think more targetable by the immune system, I think there's probably a slightly different dynamic

00:54:57.720 in that that may change things. But I would defer to the people that are doing this. I could give you

00:55:05.900 some names of people that are interested in this topic. Well, and I'm glad to hear that people are

00:55:11.620 studying this because antibiotic-resistant bacteria are a huge problem in the hospital, right? And we've

00:55:17.240 heard it, you know, again, we've heard about these super bugs. And you can't help but wonder based on

00:55:21.760 what you're saying, if a big part of the problem was we're creating them by over-treating in situations

00:55:27.880 when we don't need to, using a drug for a lot longer than we need to. Let's give people a sense

00:55:33.380 of some numbers. Because I, you know, I remember when I was in medical school, I spent a few months at

00:55:39.720 the NIH and the NCI, and I got all these amazing assignments to do. And what my favorite assignment to do

00:55:46.460 as the medical student was to look up all of the literature on cancer growth rates. And they gave

00:55:52.320 me this assignment because of my background in math. And so it was sort of like, I could go digging in

00:55:56.520 all the literature on, you know, every, you know, starting with the most trivial sort of logistic

00:56:01.740 models to the Gompertz model and all of these other models. But one of the things that really is

00:56:08.000 interesting is just how many cancer cells you need to actually even be detected, right? It's about a

00:56:16.040 billion cells in a centimeter cube. That's usually the number that's used, yeah.

00:56:21.320 Yeah, you couldn't even, there's no clinical, outside of what we now have as liquid biopsies,

00:56:26.120 there's no clinical way to even detect cancer shy of about a billion cells. And this speaks to the

00:56:32.240 non-linearity, right? To go from a million to a billion versus a billion to a trillion. Like these are,

00:56:41.660 it's hard for people to wrap their head around what's involved in those things. And both from a growth

00:56:45.920 perspective, because that's not linear and a size perspective, because you're dealing with something

00:56:51.100 that's growing with a cube power, not a square power. Where do you see this being an issue?

00:57:01.060 Because the numbers are so enormous, right? And this gets to your question of, if we talked about

00:57:08.300 the example I used earlier of there being like a million moths, well, the heterogeneity there is

00:57:14.560 already significant. But once you're talking about tens of billions of cells, by definition,

00:57:21.160 the heterogeneity is probably even more significant, correct?

00:57:24.840 Absolutely. And it's really important. I think there's two factors to think about. One is stochasticity.

00:57:31.860 If you have a small population, small changes in birth and death rates can be quite significant.

00:57:37.860 You know, small populations can go extinct based on very small changes. The other thing is what's

00:57:45.000 called the Ali effect. And that is that in a classic evolutionary model, as a population grows,

00:57:52.700 the fitness should be decreasing because you're getting toward the carrying capacity of the environment.

00:57:58.020 But Ali and others have found that it increases. And the reason is that there's an advantage in

00:58:05.660 groups. In herds, for example, I mean, you can...the obvious advantage is that they can gather

00:58:10.960 themselves to defeat predators. But they can also...there's sort of leadership issues and other

00:58:17.140 things. In tumors, there's a couple of things. One is they have to make blood vessels. And that means

00:58:22.720 that there's small groups of them have to get together and make the endogenic factors that they

00:58:28.480 then send to the blood vessels to bring a blood vessel in. It's probably not a single cell that can

00:58:33.320 do that. It's to be a loosely organized system. The development of X-cellular matrix that, you know,

00:58:38.760 the tumors...it's just not tumor cells. They make a collagen and other things that go in the...between

00:58:45.360 the cells that provide a kind of a structure for them to live on. They have to make that. And it can't

00:58:51.580 just be anything. They probably act together for that. And then they also probably act together

00:58:57.320 to defeat the immune system. They can secrete things. So there's a number of things that they

00:59:03.620 can act together on. Tumor cells, small populations, may not grow as fast as large populations. And they

00:59:12.660 may be more vulnerable to treatment. They may not be as able to deal with immunotherapy or target

00:59:18.660 therapy or even chemotherapy because they don't have the capacity to act together. So those two

00:59:27.020 things are, I think, important factors in understanding treatment. And this is a little bit

00:59:31.660 why, you know, I was talking about when, you know, there are therapies now that can take large cancers

00:59:37.000 and drive them down so that they become not detectable by CT scans, which means that their population

00:59:44.720 has to be very small. Those guys are now vulnerable to dynamics related to stochasticity and ilifax

00:59:53.720 that the larger populations are not necessarily responsive to. And one of the things that's very

00:59:59.920 interesting, if you take something like neoadjuvant therapy, where you take a big breast cancer,

01:00:04.440 for example, you make it small and then take it out, what you find is that the tumor isn't a ball that

01:00:11.100 just shrinks. So you don't, you have to start with a big ball, you get a smaller ball. In fact,

01:00:14.220 it fragments. And so what you get are little islands of tumor cells surrounded by fibrosis or

01:00:21.800 necrosis or something like that. So these guys are islands, you know, these are highly vulnerable.

01:00:27.360 And so I think this is when we should crank it up and kill it. I think tumors are vulnerable at this

01:00:34.140 point. But I think that the way we've done things, which is this dogma in oncology, you know,

01:00:40.380 continuous maximum tolerated dose until progression. And as you just said, you can't see it. If you're

01:00:47.180 waiting for progression, you won't see that until you have billions of cells. I mean, it has to be

01:00:51.580 big enough that you can image or feel. And that's at least a cc of tumor. And it's probably two or

01:00:57.340 three cc's. By that time now, when you change therapy, you're dealing with a really big population,

01:01:03.160 whereas you could have been dealing with a small fraction of it, you know, a few months earlier

01:01:08.380 when you couldn't see it. Now, the oncologists always get, well, I have to be able to see something

01:01:13.020 to be able to understand whether it's effective or not. You know, my therapy is effective. Yeah. And

01:01:18.500 that's, you know, I get that, but you're not going to be able to do that in this setting.

01:01:22.920 What's interesting is that the pediatric oncologists learned this a long time ago. This is how

01:01:26.660 they cure leukemia. They give an induction therapy, and then they immediately change to a

01:01:32.140 different therapy. So what they learned was that even after induction therapy, when there was no

01:01:38.340 apparent tumor in the bone marrow or in the blood, that if you didn't do anything, the kids would all

01:01:43.520 relapse. And so they hit it immediately with another group of drugs and then another group of

01:01:49.600 drugs. So this is first strike, second strike kind of approaches, but they're not, they can't measure

01:01:55.880 their outcome. They can only measure their outcome in the long term by saying, well, we've cured this

01:01:59.380 kid. Right. They can't say which drug was more effective in that cocktail. Right. And you'll

01:02:04.320 probably never know. There's this thing called the extinction vortex that they talk about in

01:02:09.060 extinctions, where when you've got a small population, there's typically a sequence of

01:02:14.500 perturbations. And they tend to be self-synergizing. In other words, they, the further down the population

01:02:20.180 you push, the more sensitive it is to stochastic and alle effects. And so now you can never really

01:02:27.200 tease apart those dynamics. It just, it just all goes into extinction eventually. And the goal then

01:02:33.460 is not to try to really put these things together in a really additive way, except to say that all

01:02:40.240 these guys together will tend to push this to this population to extinction. Again, it's a different

01:02:45.900 strategy and it's a little less precise than I guess, you know, we would like, but that is, you know,

01:02:53.200 the nature, I think of the, of the extinction process as we've learned it. I mean, we, it's

01:02:59.040 interesting because whenever you say extinction, everybody thinks about the dinosaurs. And I would

01:03:03.240 argue even that some of our therapy is that kind of single event, single cause extinction,

01:03:10.220 application of massive evolutionary force to the species and you wipe it out. And of course,

01:03:15.440 the problem is you also wipe out most of the land animals on earth. So it's, it's indiscriminate.

01:03:21.800 Whereas what we've learned from Anthropocene extinctions, from our own species eradicating

01:03:26.180 other species, you know, as unfortunate as that is, we've been able to study those in detail

01:03:31.100 and have learned that in fact, most extinctions are multi-cause, multi-event. It's not the dinosaurs.

01:03:37.600 They are the exception, not the rule. And so when we want to think about an Anthropocene extinction

01:03:43.680 of cancer cells, which is, which is arguably what therapy is, we should probably be thinking more

01:03:49.280 about these multi-cause approaches rather than trying to find a magic bullet that could cure the

01:03:54.800 cancer by itself because you always have these side effects. So far, we've not found a magic bullet.

01:04:01.500 Well, and to your point, the bigger issue with the dinosaur analogy is not even so much

01:04:07.460 that it's the exception and not the rule. It's the collateral damage. When there's a single strike

01:04:13.880 success, the collateral damage changes the planet. If what killed the dinosaurs when it killed them

01:04:20.920 came on our planet today, it would probably kill a lot of other things as well.

01:04:25.180 Absolutely. Sometimes when I tell people about that, you know, I said the 100 grams of tumor,

01:04:29.840 which is a, which is still a pretty small tumor, is going to have 100 billion cells, let's say,

01:04:36.260 which is far more than the human population on earth. Think about, could you do one thing? Could

01:04:41.940 you apply one event that would kill all the people on earth and nothing else? I mean, probably something

01:04:49.120 could be done, but I think that'd be very difficult. And it's a kind of a gruesome analogy, but at least

01:04:54.200 it, it kind of makes the point that the idea of a magic bullet, I think it may not be an achievable

01:04:59.340 result. What role does time play in the sequencing of this? So let's go back to the example of pediatric

01:05:05.220 oncology where you could sequence, you know, you go induction therapy and then you go with a series

01:05:12.140 of therapies to follow. So single shots, right? So one, two, three, four, five, six versus all six at

01:05:20.460 once. Let's put aside the toxicity for a moment, because a big reason you wouldn't do that is the

01:05:25.140 toxicity to the child. And a lot of these, you know, you have synergistic side effects that are

01:05:29.880 intolerable, but just thinking about this from a theoretical standpoint, would there be an advantage

01:05:34.660 or disadvantage to one of those approaches? It would certainly be reasonable to say, well,

01:05:39.780 if we've got these three different therapies, why don't we put them together and give them all at once?

01:05:43.320 And oncology has been doing this for a long time. And the results have been pretty much

01:05:47.920 uniform in that what we find is that they do better than one, you know, the tumor is controlled

01:05:55.560 for a longer period of time. But the result is still that they die of the disease, that the tumor

01:06:00.600 comes back and they die now. And you can add more, you know, four or five drugs. And what tends to

01:06:06.720 happen is that you get this diminishing return so that you just start getting more toxicity and no

01:06:11.620 increased benefit. One of the things to think about is suppose you have two drugs and your

01:06:17.700 applying it to a 10 billion population. You know, so what is the probability that there are going to

01:06:25.080 be cells present that are resistant to both? You have a big denominator, you know. So but now let's

01:06:30.660 say you have a million. So let's say you have one drug, you get it, you know, you really do a great

01:06:37.120 job. You're down to a million cells, or let's say 10 million cells. Now you add the second drug.

01:06:42.660 So now what is the probability that within this 10 million that you will have some resistance?

01:06:50.460 Knock it down again, give a third drug. Now what's the probability, let's say of a hundred thousand

01:06:55.420 cells? So Bob, I'm a bit confused by that still, because if you say we could take one drug and go

01:07:01.660 from 10 billion to 100 million, and then hit it with another drug to go from 100 million to a million,

01:07:09.980 and then a third drug to go from a million to eradication, wouldn't those be independent events

01:07:16.400 in the sense that if you just hit the 10 billion cells with all three drugs, shouldn't you still be

01:07:23.800 able to stratify that resistance pattern? What is it about sequencing those that would create

01:07:28.920 a treatment advantage? There's two things. One is you get, as you get smaller, you're both killing

01:07:34.580 the tumor cells that are sensitive. But remember that the ones that are resistant are also have to

01:07:41.960 deploy mechanisms of resistance. Ah, bingo. We're right back to that discussion. And I guess to your

01:07:48.500 point, they're going to be more fragmented and therefore have less of the group network effect.

01:07:54.420 Right. And so now when you've got a small population, subtle changes in birth and death rates

01:07:59.720 will be sufficient to drive it to extinction. That makes sense. And as I talk to people about

01:08:05.040 potentially designing this, even though there may be resistance, that you can still use the

01:08:10.740 resistance to your benefit. And some people have talked about then adding things like

01:08:16.160 environmental changes. So let's take an anti-angiogenic, so that you're now you're delivering

01:08:21.440 less blood and you're creating a more difficult substrate environment. There's people that have used

01:08:27.880 antacids. People have used, you know, a number of different, these aren't typically common medical

01:08:33.420 practice, but what individuals are doing based on their own exploration through Google and other

01:08:39.180 websites is that, and they use various metabolic things. And I think there's potentially advantage

01:08:45.540 to that, again, in combination with this in a way that makes sense. So far, most of what we've been

01:08:51.880 speaking about is really more or less one class of drug with a slightly different take on another

01:08:57.800 one, right? We've been talking about chemotherapy. These are chemicals that are aimed at disproportionately

01:09:04.200 killing cells that in theory are dividing, right? So let's go back to what you said earlier.

01:09:11.640 The thing we learned in medical school that was incorrect was cancer cells divide more quickly than

01:09:16.420 regular cells. Those of us that pay attention today know that that's not actually true at all. The

01:09:22.040 fundamental difference between a cancer cell and a non-cancer cell is the response to cell cycle

01:09:26.200 signaling. It's that a normal cell doesn't divide slower. It just, when it's told to stop dividing,

01:09:32.060 it stops dividing. Cancer cell doesn't. When it's told to stop dividing, it says,

01:09:36.700 piss off, I'm going to keep dividing. That's a very good point. I mean, that really is. And

01:09:41.240 evolutionarily what this is, is that it has a self-defined fitness function. Its proliferation

01:09:47.180 is dependent on its own properties as they interact with the environment, not on instructions from the

01:09:53.760 tissue. Normal tissue cannot evolve because its birth and death is dependent on tissue controls.

01:09:59.620 That's right. So knowing that chemotherapy, which is the earliest form of drug in the modern arsenal

01:10:07.960 against cancer, says, how can we kill a cell that is not responding to cell cycle signaling? And the

01:10:15.960 first shot across the bow is, well, let's just kill anything that's dividing because we know that

01:10:21.300 normal cells are less likely to be dividing and cancer cells are more likely to be dividing.

01:10:26.760 So let's target different cycles of the DNA proliferation phase. And that's why, of course,

01:10:33.020 most people who are getting chemotherapy have side effects to the normal tissues that relate to that.

01:10:38.980 So the mucosal ulcers, the hair loss, the skin damage, nail thinning, all of that stuff is because

01:10:46.160 those are tissues, even though they're normal, that are dividing more rapidly and they're being

01:10:50.660 smacked as part of the collateral damage. You mentioned another class of drugs, courtesy of

01:10:57.020 someone named Judah Folkman, who was really one of the first people to point out this idea that you've

01:11:02.440 raised, which is cancer cells have another challenge that they face, which is they have to get the blood

01:11:08.320 vessels to bring in the substrate, the glucose that they need to replicate, make energy and make

01:11:16.140 building blocks to make more cells. And what if we target those things, these vascular endothelial

01:11:22.900 growth factors? And those drugs haven't turned out to be very successful, by the way. So they're

01:11:28.120 billion dollar blockbuster drugs, you know, Avastin probably the first, if I recall. And they've

01:11:34.400 extended median survival like a few months here and there, but they really haven't been blockbuster

01:11:38.800 drugs. We haven't talked about immunotherapy, right? So we haven't talked about the Keytrudas of

01:11:45.060 the world, how they fit into that. But it seems to me, and there's others, of course, right? We can

01:11:51.300 talk about the few cancers where a single gene mutation exists and you can target it with like a

01:11:57.980 LEVAC or something like that. But it seems that you could use selective chemotherapies and

01:12:06.080 target on top of that anti-VEGF. And the real question then becomes timing, right? How do you

01:12:14.460 cycle those on and off to always be maximizing the gap in fitness? So how do you maximize that fitness

01:12:24.060 ratio? By the way, did your model predict that there's something specific about the number seven?

01:12:29.760 Because if I recall, you said your model thought you would end up between two and three. You actually

01:12:34.300 ended up at seven. In retrospect, does the model now tell you seven is the tipping point?

01:12:39.260 It's somewhere around there. We've looked at that, at the transition as you go from

01:12:43.540 zero to two to three to seven. It's in that range. I don't know exactly where it is. And it's

01:12:50.560 certainly going to vary from tumor to tumor, but it's in that, it's certainly,

01:12:54.760 I would guess at the five-ish range. Okay. And so going back to my question,

01:13:01.260 do you think, for example, the anti-VEGF drugs play a critical role in this when the tumor is

01:13:06.800 scattered? The way I'm thinking about it, right? I've anthropomorphized everything you've said

01:13:11.440 into a society, right? Which is the United States is 330 million people. It's a mighty nation.

01:13:18.000 And if anybody attacks us in theory, we can mount a response in a coordinated fashion,

01:13:24.560 et cetera, et cetera, et cetera. And this analogy may be too stupid by the time it comes out of my

01:13:29.840 mouth, but basically if you decimated the population by 90% and there were only 33 million people left,

01:13:37.940 presumably we would no longer continue to be the United States of America. We would be a whole bunch

01:13:43.340 of disparate tribes in total chaos. Therefore, we'd be a heck of a lot more susceptible. Is that

01:13:49.640 a fair assessment? Yes. And that's the, in the military, they talk about defeating your enemy in

01:13:54.580 detail, meaning that if they're fragmented, you don't have to deal with the whole population. You

01:14:00.440 can just do it one at a time. And yes, I think those are the case. That's the case. And when you look

01:14:06.240 at tumors after neoadjuvant therapy, these small little populations look like they're, you know,

01:14:12.340 a hundred, maybe a thousand cells, not millions of cells. And they're widely separated. You know,

01:14:19.480 they're not sending cells back and forth very much, if at all. So it's a very different dynamic

01:14:25.020 than these continuous tumors that, you know, where if you take, if you, you know, chop out this corner

01:14:31.220 of the tumor for some reason and all those cells die, they, everything just sort of moves in there

01:14:35.460 and eventually it repopulates. You don't have that luxury in these fragmented populations. And

01:14:41.780 that's one of the major components of Anthropocene extinctions. The Heathhand, for example, which was

01:14:48.900 all up and down the Northeast coast when the European settlers arrived, was finally just limited

01:14:56.260 to Martha's Island. And although it was protected by the humans there, a series of things happened.

01:15:03.160 A couple of bad winters, there was a fire in one, in part of their habitat, and there was a disease

01:15:09.740 developed. And although the humans protected it so that its population grew considerably, it was still

01:15:17.540 in only Martha's Vineyard. And these perturbations just essentially went to that, you know, it died.

01:15:23.540 Again, it was at the extinction vortex, several things. And you can't point to any one of them

01:15:28.680 as the cause, but, but together they wiped it out. You know, if, if they had had another colony,

01:15:35.000 somewhere else, you know, they would still be alive, but that just didn't happen. So.

01:15:39.960 So where do you think immunotherapy fits into the toolkit here, given that it's,

01:15:46.480 it tends to be more binary than chemotherapy, right? So immunotherapy,

01:15:50.740 I'm making this up because it varies by histology, but let's just say you take people

01:15:56.500 in the best case scenario to people that have checkpoint mutations, right? So someone that has

01:16:01.440 a PD-1 mutation, the response rate to those people for an anti-PD-1 drug like Keytruda is very high

01:16:09.400 and very durable. I mean, far more than anything you see with chemotherapy. Conversely, if they don't

01:16:15.440 have the PD-1 mutation, the drug is useless. So what does that tell us about the opportunity to use

01:16:22.780 immune-based therapies? And by the way, we borrow the term adaptive therapy again here, right? So you

01:16:27.500 have adaptive therapies within immunotherapy, which are not to be confused with what you're

01:16:31.700 describing as adaptive therapy. Yeah, I think it's going to be a critical component of all this. And

01:16:36.540 as you say, sometimes immunotherapy can stand on its own. It's the closest thing

01:16:41.140 to a magic bullet that I think we have. The group I work with tends to view immunotherapy,

01:16:48.300 with the exception of those unusually responsive tumors, as your closer. And this is our baseball

01:16:53.900 analogy that when you get the population that's small, we think that alle effects are particularly

01:17:00.180 important in immune response, in response to the immune system.

01:17:03.480 Meaning evasion from the immune system. Right. And so bringing the immune system in

01:17:09.540 when you've got the tumor on the mat, you know, it's small, fragmented, I think that will be the

01:17:17.000 most effective closer that we have to essentially wipe it out. I could be wrong, but that in theory,

01:17:23.300 this is the one you bring in to win the game. And by the way, that could be going back to Gen 1

01:17:28.440 immunotherapy, which is interleukin-2. Interleukin-2 had about maybe a 10 to 20%

01:17:34.940 response rate in melanoma, a particularly immune-sensitive cancer. But that was treating

01:17:40.280 patients with full-blown metastatic melanoma. Maybe it would look very different in patients

01:17:46.220 who were NED, or who are, you know, visibly without disease. And there are some, I know that

01:17:52.640 there have been some experiments that have suggested that cells that survive chemotherapy

01:17:57.440 are more vulnerable to immunotherapy. And ultimately, that would be the game that you'd

01:18:03.240 want to play here, is that you, what you want to do is apply a therapy, and you know that the

01:18:10.620 resistance strategy might be the following thing. You know, this is how it's going to evolve to become

01:18:17.120 resistant, and then attack the Achilles heel that's exposed by this immune response. We had a person here

01:18:25.820 do a, that's kind of the adverse of this now. They did a P53 vaccine in lung cancer patients that were,

01:18:33.560 had been treated with multiple different things. And it didn't really get an effect. I mean,

01:18:38.820 there was minimal efficacy. One patient got a partial response. None of the others did, although they did

01:18:44.620 generate immune cells to the vaccine. They then went on and gave them chemotherapy. And the response rate

01:18:50.960 to the chemotherapy was 60 to 70 percent, astonishing number for patients at that stage

01:18:57.760 in their disease. It should have been less than 5 percent. And the patients that had the best immune

01:19:02.860 response to the vaccine were the ones that were most responsive to therapy, suggesting that those that

01:19:09.900 responded, you know, they had, they, they developed adaptive strategies. And those adaptive strategies

01:19:16.640 then made them more vulnerable to toxicity. I mean, it's appealing to think that maybe they

01:19:22.160 down-regulated P53 so that they were not expressing the target. And of course, P53 is important in

01:19:30.580 survival mechanisms and that sort of thing. So perhaps that was a mechanism. I'm just speculating,

01:19:36.600 of course. But that's the kind of thinking that I would like to apply here so that we're not

01:19:42.480 isolating, you do this therapy and then you do this therapy and you do this therapy, all of them in

01:19:48.900 isolation, but start to put strategic strings of therapy together where the adaptive strategies that

01:19:56.480 you're, you're going to select for with one become vulnerable to two and three. And so that you're putting

01:20:02.460 them together in a strategic kind of thoughtful way rather than just kind of putting them together

01:20:09.180 in a kind of haphazard or intuitive way. And again, that's where math models can, I think, can be very

01:20:14.360 helpful.

01:20:15.160 No, that, that, that actually makes a ton of sense. I want to talk about another feature of, of cancer,

01:20:20.580 which is what I call the source sync trade-off. I'm sure you've thought a lot about this. I endlessly

01:20:27.720 thought about this and, and really never came up with a great insight. Let's start from first principles.

01:20:33.960 There aren't that many cancers that can kill you without spreading. Let's list the ones

01:20:39.080 that can. Brain cancer can kill you without spreading. So glioblastoma multiforme can kill

01:20:46.160 you and it doesn't leave the brain. Primary hepatic cancer can kill you without spreading.

01:20:51.840 Hepatocellular carcinoma can kill you. I suppose lung cancer can kill you without spreading though

01:20:57.660 it's less common. Am I missing an example? Is there some other cancer that can kill you shy of

01:21:03.860 metastasizing? I think you've, you've, for all intents and purposes, yes. Okay. You're right.

01:21:09.820 Okay. So that's principle number one. Then let's go to observation number two or observation number

01:21:15.980 one, I suppose. So there are certain cancers that are very deadly that leave tissues that other cancers

01:21:23.740 never come to. So breast cancer, prostate cancer, pancreatic cancer, colon cancer. These are deadly

01:21:34.760 cancers, which come from an environment that are not especially hospitable to cancers. Because when

01:21:42.700 a woman gets breast cancer, the first thing that cancer wants to do is get out and kill her by going

01:21:48.280 to her bones, her brain, her lungs, her liver. Pancreatic cancer virtually always wants to go to

01:21:54.800 the liver and that's where it kills. Colon cancer wants to go to the lungs, wants to go to the liver

01:21:59.980 and wants to go to the brain sometimes. And that's where it kills. But rarely does a cancer go to the

01:22:06.640 breast or go to the pancreas or go to the colon or go to the prostate. Prostate virtually always wants

01:22:14.380 to kill you by going to the bones. So do you see where I'm going with this? Like you have this,

01:22:20.600 and I could go on for hours with all of my observations, right? Like the brain is another

01:22:24.120 one, right? It's an enormously attractive place for cancers to go to, but a primary cancer from the

01:22:30.820 brain never seems to leave. Now, theoretically, there's an exception to that. I suppose there are

01:22:36.220 some vascular tumors of the brain that can leave, but they're not really brain parenchymal tumors.

01:22:40.600 Do you have an insight into what explains this? And more importantly, does the answer to that

01:22:46.940 question or the inference of an answer to that question offer an insight into the environment

01:22:53.400 in which tumors thrive and therefore what we might be able to do about it?

01:22:58.780 The short answer is I don't know. Let me say what I'm certain of, and that is this is not

01:23:04.560 a planned event. We tend to give cancer cells this sort of anthropomorphic kind of, you know,

01:23:12.420 they are, first they go out and prepare a metastatic site, and then they send their cells out to it.

01:23:19.600 That doesn't happen. An evolving population can never adapt to conditions it has not seen before.

01:23:26.720 It cannot plan to make metastases, and it certainly can't plan to feather its nest, you know, in some

01:23:36.100 distant site before sending out its scouts. That's not happening. But we know in nature that we see

01:23:45.120 introductions all the time, and we know from nature that species introductions are sometimes successful

01:23:53.080 and sometimes not, and it has to do with the way their phenotype interacts with the local adaptive

01:23:58.660 landscape, and can they adjust or not. And we know that sometimes it takes several introductions

01:24:04.620 before one occurs. Sometimes it never happens. Sometimes they can expand and then die out.

01:24:11.660 Again, they're just basic rules that they have to, you know, live by. So, for example, you can see

01:24:17.880 species coming out of the Amazon. They get onto, the Amazon collects trees and things that are floating

01:24:24.580 down, the animals are on it, and so they go out. And then anything that's in front of the Amazon is

01:24:29.620 going to be receiving more of these, so are more likely to see a metastatic monkey or something like

01:24:37.560 that from the middle of the Amazon. So, the pancreas spills into the portal vein, which then goes into the

01:24:44.240 liver, and so it's delivering a lot of its cells to the liver. So, it makes sense then that it tends

01:24:50.440 to metastasize there, if only because it's just sending a lot of cells there.

01:24:54.800 Right. The pancreas makes sense, but how do we make sense of the breast? How do we make sense of

01:24:58.760 the prostate going to the bone disproportionately? I think we just suggest general principles,

01:25:03.800 that there's something about some of the breast cancer cells seem to be able to set up shop in

01:25:08.940 bones, some of them in the lung, and for reasons that are not yet clear. But you mentioned actually

01:25:15.680 something that's, I think, important, and source sink. And I think that's a dynamic, and that's

01:25:21.080 something that we've been talking about recently, source habitats, sink habitats. The idea that you've

01:25:26.560 got habitats that have very good blood supply, that they produce a lot of cells, and then now they're

01:25:32.600 producing too many cells for the spatial environment, and those cells have to go out. And one can think

01:25:39.340 of those within a cell so that you've got areas with poor blood flow and areas of good blood flow

01:25:43.540 next to another, and you can imagine cells migrating between them. And this can set up dynamics that are

01:25:49.020 very interesting. But you could also think about a breast cancer cell that's got, let's say,

01:25:54.160 is near a blood vessel, and as you might expect it, so now it's getting big, there's too many cells,

01:25:59.860 they crowd into the blood vessels, it sends them out. And so now you've got perhaps sink habitats

01:26:04.780 somewhere else, and there may be some coupling that goes on in ways that, honestly, I don't

01:26:09.480 understand. But I think that we can sort of make up rules for that. But for sure, what we know is that

01:26:16.480 the limiting factor in this is the dynamics at the metastatic site, that if you inject cancer cells

01:26:23.800 into mice, that nearly all of them die at the metastatic sites, only a very small percentage of them

01:26:29.280 that will form even a few cells, and a smaller percentage of those that ultimately form more

01:26:35.020 cells that form a population. But I think this, again, is that small population dynamics that we've

01:26:40.820 talked about. There's stochastic effects, alle effects. You know, there's a lot of statistical

01:26:46.980 problems with going from a single cell to a cancer that's significant. We are very lucky for that,

01:26:54.260 because we know that human cancers are frequently dumping millions of cells into the blood. And

01:27:02.040 yet metastatic sites, metastases are relatively rare when you think that. You know, people with

01:27:08.360 early stage lung cancer, you know, you can find cancer cells in their bone marrow, and yet they

01:27:14.200 don't get bone marrow metastases. Breast cancer is the same kind of thing. You can do bone marrow biopsies

01:27:20.480 on women getting mastectomies for apparently localized disease. And maybe 30 to 40% of them

01:27:25.700 will have breast cancer cells in their bone marrow. And yet all of them do not develop

01:27:30.500 metastatic disease in the bones, thank goodness, because it seems like for a variety of reasons.

01:27:36.680 And how much of that do you attribute to the alle effects and the stochastic variability that says,

01:27:43.340 look, they're just not going to be fit enough to take up residence there,

01:27:46.300 versus some other inherent principle of like the genetic robustness of the tumor itself. Because

01:27:54.460 a lot of those women may go on to get adjuvant therapy, and then it becomes a question of how

01:27:59.560 successful was the adjuvant therapy? Yes, exactly. And I think that's where if you look at breast

01:28:05.840 cancer, for example, what we know is that adjuvant therapy will reduce the probability of metastatic

01:28:12.440 disease, but not eliminate it. It's at best a small effect. And we need to do better with that.

01:28:18.820 It's a great point, because how do we treat breast cancer with adjuvant therapy? Well, we give them

01:28:23.280 platinum or some drug for some period of time, and that's it. Well, why don't we give, you know,

01:28:29.340 a sequence of drugs? Why don't we take advantage of, because we know for sure we're dealing with

01:28:34.380 small populations, and we know for sure that we can cause them to go extinct at least some of the time.

01:28:40.740 So why don't we optimize how we do this instead of simply pick up a drug from the shelf and administer

01:28:49.960 it continuously, you know, for six months? You know, this is another good example of where I think we've

01:28:56.560 not thought through the eco-evolutionary dynamics of what we're trying to treat.

01:29:01.260 Yeah, it's basically a paint-by-numbers approach, right? Which is paint-by-numbers is we're just going

01:29:05.840 to do it this way versus I'm going to think through this strategically. We have a pretty good sense that

01:29:11.320 the more cancer cells a person has in their body, the more mutations they have. So the difference

01:29:17.540 between a billion cells, i.e. one cc of cancer, versus a trillion cells, which is like an almost

01:29:25.260 fatal load of cancer, is an enormous number of mutations. And therefore, it's not only bad from

01:29:32.820 the mass effect, but also from now the fitness of it, right? You have so, you just have such a

01:29:40.500 genetically, a bad genetically diverse population. So you have enormous heterogeneity within the

01:29:48.700 genetic badness, for lack of a smarter word.

01:29:52.900 Can I add that you also have diversity of their ecosystem?

01:29:57.960 Yeah, yeah, yeah. No, no, no. But let's do that. Yeah. So let's talk about these two indices of

01:30:02.720 eco and evo indices, which you've written about in a very recent paper. So explain for folks what

01:30:07.640 these mean, because these are, they're not the easiest concepts to get your head around sometimes.

01:30:12.360 So when you talk to a cancer biologist about evolution, they say it's mutation selection.

01:30:19.060 You know, you get a random mutation and that the cancer cells accumulate these mutations.

01:30:23.840 Occasionally, they provide a benefit, and that's what causes the tumor to expand.

01:30:29.780 The problem with that approach is that it assumes that the environment is stable.

01:30:34.540 They're all dealing with the same environment. Now, the alternative is to say, and I think the

01:30:39.820 more evolutionary appropriate thing to say is that there is tremendous variation in blood flow

01:30:44.620 and other factors within the tumor. And as the tumor goes into these areas of, say, low blood flow,

01:30:51.300 the environment is entirely different. And so the environment is applying selection forces

01:30:57.200 that are totally different from, let's say, down the road, where the, it's, you know, the blood

01:31:02.400 vessels are really good. And the edge of the tumor, these cells are competing with the normal cells.

01:31:08.460 Internally, they're competing with other tumor cells. So, again, entirely different environments.

01:31:13.680 And so a more evolutionary model, in my opinion, is to say that the different environments within the

01:31:20.440 tumor are the ones that give rise to different phenotypes, which in turn give rise to different

01:31:26.860 genotypes. So that in this case, the genes aren't causing the evolution. The genes are the consequences

01:31:33.280 of evolution. One of the ways to think about this is to think about whether a modern-day Darwin

01:31:39.320 sitting on the beagle, you know, in the, in the, in the ship with his microwave machine. He's got all

01:31:45.400 these fancy molecular biology equipment. And sailors just brought random samples of finches to this

01:31:53.060 modern Darwin. They ground them up, put them through the machine, got billions and billions and billions

01:31:57.840 of data points. Could they have written origin of the species from that data? And I think the answer is

01:32:04.080 no. And I think that the reason is that what Darwin saw was that the beak of the bird and the seed

01:32:10.960 matched. There's a morphologic matching that makes common sense. This isn't magic. It's just that beak's

01:32:17.780 got to be bigger to, you know, pick up that seed. He paired phenotype to environment. To environment,

01:32:22.980 to environmental section, right? That's the logical way that evolution occurs. The genes are kind of

01:32:29.240 downstream of that. The, you know, multiple genes, you know, the beak could be under the control of,

01:32:34.760 you know, a few genes. And, but how would you distinguish these genetic changes and say, well,

01:32:41.020 there's the beak there and, oh, there must be, that means there must be a big seed there.

01:32:44.720 Well, isn't it safe to say we might not even know? Like, isn't it safe to say that if you took

01:32:50.000 the two closest finches, the one that went from the beak like this to the beak like this,

01:32:56.120 and you sequenced them both, you might not be able to extract which genetic differences between

01:33:01.680 those accounted for the beak. In fact, I'd be surprised if you could.

01:33:04.700 No, I completely agree. And I think that when we talk about cancer as a disease of the genes,

01:33:12.100 it's more complicated than that. And I think that the intense effort to characterize the genetics of

01:33:20.080 tumors, you know, has had some benefit, but it also has some disadvantages because you lose

01:33:25.680 that morphologic matching of the environment and the morphology of the cell, which is more common

01:33:33.400 sense. And I think in the long term helps us understand it better. I mean, one of the questions

01:33:38.620 one of my colleagues likes to ask is, who's an evolution biologist, how many niches do you think

01:33:44.000 are present in a cancer and how many species are present? I mean, certainly not one, virtually never

01:33:49.760 is it going to be one. But is it nine? Is it a hundred? Is it a million? So when we think about

01:33:56.920 tumors, a clade, it's starting from a single cell or small number, but it's speciating all over the

01:34:02.600 place as it gets in different environments. But the speciation isn't magic, and it's not just due to a

01:34:07.800 random mutation. It's because there are environmental variations. And trying to understand that, I think,

01:34:13.320 would help us a lot in terms of really getting a handle on what the cancer is doing. And the

01:34:20.400 reverse of that is, I mean, so take cancer cells out of a patient, put them in a dish, and now you

01:34:27.180 start to grow them in culture. Well, they're no longer being attacked by the immune system. They no

01:34:32.480 longer have to worry about androgenesis. I mean, it's an entirely different...

01:34:35.960 Right. They don't have the edge effects that you described earlier, where some of them have a

01:34:39.960 selective advantage vis-a-vis access to substrate, whereas others are more competing with the tumors

01:34:45.740 at the inside. They're subjected to entirely different environmental selection forces. Now we

01:34:52.020 say we take those tumors out, we do molecular biology studies, and we say, oh, well, this came from a lung

01:34:58.840 cancer. That means lung cancer must interact with this to do this, when in fact we're dealing with

01:35:04.660 cancer cells that have evolved far past. You know, they're far different from the cancer cells that

01:35:11.320 were present. That makes up an enormous amount of the cancer literature right now. And I don't know

01:35:16.600 that that's useful. I don't know that you can extrapolate from those results to anything that goes

01:35:23.040 on in vivo. And it's also, I think, has led to very confusing literature, at least in my opinion.

01:35:28.680 You take genes that have been extensively evaluated, and it's like, they could do anything. You know,

01:35:34.660 according to the literature, they have any function you can name. They get found everywhere. So that's

01:35:40.460 becoming non-useful. If you can't really sort of pin down what its role is in vivo, in a patient,

01:35:48.080 you've got a lot of literature, but I don't know that you have a lot of knowledge.

01:35:51.500 How does this fit into the EVO index? How do you, which basically looks at the diversity

01:35:55.880 versus genetic change over time, how do you then use that to clinically make a decision?

01:36:02.140 I don't know. I mean, in dealing with a very heterogeneous population, and I think this is

01:36:07.740 your question, it does, as you increase the heterogeneity, does that increase the, let's say,

01:36:13.120 its resistance to therapy or the likelihood? And I think that's probably true. But I don't know how

01:36:18.680 that can be taken into account right now in a clinical setting. And I'm sure that someone smarter

01:36:23.960 than me is going to figure this out. So in other words, we can't do something as simple as biopsy

01:36:30.460 patients to get a sense of drug resistance versus non-resistance because it immediately takes them

01:36:37.560 out of the environment in which it matters. And we get back to this problem of, well, we can't really

01:36:42.920 do this in vitro. The nice thing, I mean, you get information about the tumor, but those cells are dead.

01:36:49.720 What's in the tumor is now going to rapidly diverge from there because the minute you touch it with

01:36:55.800 therapy or some other perturbation, it's evolving and changing very rapidly. So again, it's like the

01:37:03.940 hurricane going on further and further. Whatever the data you got on day one is important for day two,

01:37:09.800 but becomes less important on day three and day four and day five, and ultimately then having nothing

01:37:15.220 to do with, you know, having no predictive capacity at all for the hurricane. So I think one of the

01:37:22.560 things that I think we have to be careful about is how we use data. I mean, there's a lot of interest

01:37:26.880 in circulating DNA and circulating tumor cells and that sort of thing. But there's very little

01:37:32.580 information about where they come from exactly. Is the DNA coming from cells that have died because

01:37:37.700 they're less fit? So those are the losers in the evolution game. So we do not care about those.

01:37:43.440 Or are they the winners? You know, and in that case, we do care about them. And how are they

01:37:48.700 representative of all the different range of cells? Is the distribution of DNA representative

01:37:54.880 of the distribution of populations? Or I think that's highly unlikely. And so I think that we do

01:38:00.320 have to be thinking about using imaging as a way to look at the intratumal evolution over time.

01:38:07.660 It's a non-destructive way to do it. But we have to be able to take the macroscopic scale images that

01:38:14.520 we can get from radiologic studies and bridge the scale to a microscopic level that can tell us about

01:38:22.300 what's going on at the cellular and molecular level. And what kind of tool could do that?

01:38:28.960 Landscape ecologists have developed technology where they take satellite images and they can look at them

01:38:36.260 and say, and again, this is simplifying things quite a bit, but they can say, well, can we generate a

01:38:42.920 species map from these high-level images? And the way they tend to do it is that they look for habitats.

01:38:50.800 You know, they identify distinctive areas within the image. Usually there's five types. Every part of

01:38:58.060 the image is one of these five, let's say. So instead of tramping through this entire state or however,

01:39:04.160 you know, large area, you just tramp through one of the habitats and say, well, what are the species

01:39:09.320 distribution here? And then you extrapolate. I think that those kinds of things, and again,

01:39:14.980 part of the reason is that in the images, we see areas that have good blood flow, poor blood flow.

01:39:20.580 We might even see areas of temporal variations in blood flow. We can see edema. We can see a lot of

01:39:26.020 different things. We may be able to define habitats. And again, not, we won't see the cells,

01:39:31.560 but we can see them indirectly in the sense that we know that, for example, in an area where the

01:39:36.760 blood flow is varying frequently, that there's certain types of phenotypes that are going to be

01:39:43.080 adapting to that environment. And so we can at least make that extrapolation and make those kinds

01:39:48.240 of suggestions. It may be we can do better than that and even start to think about, you know,

01:39:53.680 molecular properties of the cells that are present, at least in a general way. It's not perfect,

01:39:58.620 but it's at least another mechanism, perhaps in combination with blood circulating tumor cells and

01:40:04.760 or cell-free DNA and cell-free DNA, all those things combining with that. I think probably

01:40:11.480 neither alone is going to be sufficient, but together they may be sufficient that you can

01:40:17.440 understand intratumal evolution during therapy, which is really a key piece of clinical data that

01:40:22.420 we currently cannot get. What inference do you draw from your work with respect to the importance or

01:40:30.040 lack thereof of early, early screening? In other words, do you now have a point of view one way or

01:40:37.760 the other from the mainstream view, which truthfully is that screening is kind of important, but probably

01:40:43.340 not that important, right? I mean, there's only five cancers for which probably the American Cancer

01:40:48.440 Society would even offer. Certainly the U.S. Task Force on Preventative Services would even offer a

01:40:53.880 point of view on screening. Prostate, we've basically thrown our hands up and said, eh, talk to your

01:40:58.740 doctor. So we have a point of view on lung cancer and smokers, mammography, cervical cancer, and colon

01:41:06.320 cancer. But given that we now have many more tools to go looking for cancer early, whether it be liquid

01:41:12.380 biopsies and cell-free DNA or far, far better imaging studies, does your work point you one way

01:41:18.120 or the other towards having a better shot at treating cancer by getting it early? It would make sense.

01:41:23.180 Again, the smaller the population, the more likely you could cause extinction. But that said, I'm not,

01:41:28.720 I don't think really qualified to comment on the screening. I am very supportive of screening. It just

01:41:36.220 intuitively makes sense. But that's just from a personal point of view. I'm not based on any work

01:41:42.260 that I've done. So when I hear what you were saying, it makes me think you should be doubling

01:41:48.000 down on screening, right? It makes me think that notwithstanding the challenges of screening, which

01:41:53.520 are both economic and the psychological toll of false positives, which invariably happen the earlier

01:41:59.060 and earlier you look for a cancer, the trade-off is those effects that we've described, which is

01:42:05.940 you are going after a group of hunter-gatherer colonies rather than the United States of America,

01:42:10.600 right? You have a much easier chance and they've had far less time to accumulate mutations and they

01:42:17.100 are far less interconnected. Far more vulnerable. Yeah, exactly. Why did you select prostate cancer

01:42:23.100 for your trial? Was it simply because you had a great biomarker in PSA or was there anything else

01:42:27.440 about the biology of prostate cancer that attracted you? Two things was that it had a great biomarker and

01:42:33.280 it had a very brave oncologist willing to do it. And you should not minimize the courage of the latter

01:42:38.980 because it does go against the grain and the oncologists have to swim against the stream to do

01:42:46.300 this. And many of them are really not willing to do it. So I salute Jin Song Zhang, who was the very

01:42:52.340 brave oncologist that ran the trial and did it very well. Well, given the success of that trial now,

01:42:58.260 has that increased the appetite of others to consider doing this either in a larger scale with randomization

01:43:05.060 for prostate cancer and or for other histologies? Minimally. There was a philanthropic fund in

01:43:11.020 Europe that wants to fund a phase three trial in prostate cancer. So that's really good. And I think

01:43:18.120 that's really important. But I would say that at the same time, if anything, I get the feeling there's

01:43:24.060 more resistance. I think just everybody's got their backup. It's interesting, the article that I mentioned

01:43:29.640 about showing how you can eliminate the resistant cells. I got this morning, actually got the 10th

01:43:36.920 rejection for that article. I mean, it just, people just don't want to hear it. So I don't get the sense

01:43:42.960 that there's any greater interest in that sense. If funding were available, and more importantly than

01:43:49.200 funding, or at least equally important, would be oncologist collaboration. What other experiments would

01:43:55.140 you want to do? Well, I think there's some low-hanging fruit. I think ovarian cancer is one that could be

01:44:00.680 treated using this technique very successfully. Again, they have a nice serum marker. I think that

01:44:07.340 small cell lung cancer, almost universally fatal disease, but one that is responsive extremely well

01:44:13.220 to initial therapy. So, you know, any tumor that responds well to initial therapy to the point that,

01:44:18.960 you know, you virtually cannot see any tumor is one that in theory could be, I think we could

01:44:24.920 eradicate, and then initial prostate cancer therapy. So pre-metastatic?

01:44:30.780 No, metastatic. Oh, metastatic. Okay.

01:44:32.640 Men that present with metastatic prostate cancer, they're treated with androgen deprivation therapy,

01:44:37.980 and the PSA becomes normal or undetectable in 90 plus percent of these men. And of course,

01:44:45.740 what do we do? We just keep treating with androgen deprivation therapy.

01:44:48.820 Which, by the way, says nothing about the metabolic damage and the total metabolic derangement

01:44:54.820 it brings on those men. So if they don't die from prostate cancer, they're going to die from

01:44:59.060 diabetes and the complications of fatty liver disease.

01:45:03.000 Yeah. I mean, they hate the therapy. So I think as soon as it normalizes, we should

01:45:08.340 use additional therapies. Hit it again. You know, this is the knife fight again. And that's a very

01:45:13.760 common disease and a very common scenario. And it's interesting that I cannot get anyone,

01:45:20.040 any oncologist, to be interested in doing that. To me, that's very low-hanging fruit that we should

01:45:28.360 certainly try. Or we could try the cycling, but with the idea that we drive the resistant cells to

01:45:35.360 extinction, also probably fairly easy to do, perhaps even easier to do. But with the idea that we get to

01:45:41.800 the point where these men don't have to keep taking this, the energy deprivation therapy. I mean,

01:45:45.260 it's funny because you're exactly right. I mean, I think there tends to be the sense of,

01:45:49.900 well, it's not really cancer therapy, and it's certainly not like chemotherapy.

01:45:54.640 I find it to be worse, truthfully, for the patients.

01:45:57.240 Men are miserable. And we should not minimize the effect of this, which is very significant.

01:46:02.660 And it's a pretty common problem. So I think that there are some opportunities. I don't think,

01:46:08.560 to be honest, I don't think this will be something that I'll see in my lifetime. But I do hope that,

01:46:14.760 you know, the next generation of oncologists and cancer biologists will try to at least bring this

01:46:21.780 another step forward. Is there a cancer that you would stay away from at the early stage? In other

01:46:27.820 words, you obviously want to start with the ones where you think you have the highest likelihood of

01:46:31.280 success. What would you put at the other end of that spectrum where you think, you know,

01:46:34.320 that's going to be a really tough problem to solve? I think the toughest problem I can think

01:46:39.580 of is glioblastoma, one that you mentioned before. I just don't understand how, I mean,

01:46:46.520 every other tumor virtually, the problem is not local control. If you control locally, you cure it.

01:46:52.920 It's a mastatic disease. GBM, you can't control it. It bothers me that there's something we're

01:47:00.620 missing in this that I don't understand. But it feels like, first of all, it's an intimidating

01:47:06.120 cancer and I would never try to get involved with it if, I mean, I'd love to be able to make an impact

01:47:12.420 to help people with the disease. But it's such a difficult problem that I would be loathe to do

01:47:18.960 that unless someone came up with a really good idea for treating it. I mean, I think part of the

01:47:23.980 problem is there's the, quote unquote, the simple mass effect problem, right? Which is anything in the

01:47:29.180 brain is problematic due to the inability of the organ to absorb growth. I think a second problem

01:47:35.900 with GBM is the ubiquity of radiation that is understandably used to treat it, but it introduces

01:47:43.060 mutagenesis probably at an even greater rate. So now you have two layers. I think it puts a bit of an

01:47:50.400 accelerator on its ability to out-compete its environment. It just introduces a new genetic

01:47:57.480 heterogeneity. I don't know if that's right. I mean, that's just a theoretical argument. I don't

01:48:01.420 know that that's been documented. Of course, the brain is inaccessible and so it's hard to see what

01:48:05.960 goes on. But what we know is that, you know, you hammer this tumor. I mean, you take it out,

01:48:12.940 debulk it, most of it, give it radiation therapy, chemotherapy. I mean, very little can survive

01:48:18.600 that. And yet it does. And the question is, is it because, as some people think that tumor cells on the

01:48:26.380 periphery sort of grow back into it? Or is it because there's resistant cells that emerge that

01:48:32.260 are so tough that you can virtually hit them with a hammer and they just shrug it off? One of the

01:48:39.600 things I had suggested at one point, maybe we could do radiation first, then do the surgery and see if

01:48:46.300 you can learn about how the cells change during radiation therapy that could give you some idea

01:48:52.520 about whether they're evolving resistance, whether that's the major factor or not. But it's very hard

01:48:57.720 to convince anybody to do it. And the surgeons, of course, are loathe to work on anything that's

01:49:03.300 had radiation before it. But it's interesting because the neurosurgeon said, well, you could

01:49:07.240 probably do it. But nobody wants to do it. Well, everybody knows that's not what you're supposed to

01:49:12.600 do. So it's never gone anywhere. And of course, I've not pushed it because I would be afraid of

01:49:18.520 causing harm. What about other bad actors? I mean, I always put cancers in the category of

01:49:24.000 there's the cancers that give cancer a bad name, like GBM is one of them. Pancreatic adeno is

01:49:29.220 another. What would be your level of optimism or pessimism around pancreatic adenocarcinoma?

01:49:35.380 Again, I've sort of worked with some people on this. I think there's still there's opportunities.

01:49:40.680 Again, I don't think we understand enough about the biology. So for example, classically,

01:49:45.180 pancreatic cancers have a lot of fibrosis associated with it. One question is, is that

01:49:51.000 fibrosis a host response? Or is it a tumor adaptive strategy?

01:49:56.720 Evasion. Yeah, yeah.

01:49:58.300 Now, nobody really seems to know that. So little things like that. So should we promote fibroblasts?

01:50:03.640 I mean, you know, classically, we try to reduce the fibroblasts because there's this idea that you can

01:50:09.880 get more therapy in if you do that. But suppose it's a host response. Maybe we should be giving

01:50:15.600 fibroblasts growth factor to them. And maybe they're competing against each other for space.

01:50:21.420 And if you just promote the fibroblasts a little bit, they'll kill off the tumor cell for you.

01:50:25.400 I'm just saying that with every tumor, we could ask questions because we often don't know really

01:50:31.140 basic things about the underlying eco-evolution dynamics. And if we could learn more about them,

01:50:38.780 I think we could have some different solutions, some different strategies. I will point out that

01:50:44.720 another brave pediatric oncologist that I work with, Damon Reed, has started a trial for kids with

01:50:52.040 metastatic sarcoma. And as for giving cancer a bad name, this is one of those tumors that responds

01:50:59.560 very well. This is a rhabdomyosarcoma. It responds very well to chemotherapy and then comes back and

01:51:05.360 they die. And of course, these are teenagers, young adults. I can't think of anything more tragic

01:51:10.280 short of younger children dying. And so he is actually trying an extinction therapy protocol

01:51:16.460 in a trial. So I just wanted to give him some recognition for doing this. And God, I would love

01:51:24.120 to help these kids. And there's no doubt, Bob, that there are going to be people listening to this

01:51:28.600 who either themselves or their loved ones are progressing through therapy and basically looking

01:51:36.940 down the barrel of no options. And so let's say someone's in that boat and they hear this and they

01:51:42.580 say, well, look, I'm basically being told by my oncologist I've progressed through all my therapy or

01:51:47.460 the term that oncologists use, that some oncologists use that I don't particularly find fond is you failed

01:51:53.040 therapy. I hate that expression. Yeah. Now let's say they're listening to this and they say,

01:51:57.280 I want to find a doctor who can help me try this therapy. What would their options be?

01:52:02.060 Well, you know, you have to find an oncologist willing to do this. And there are some who will

01:52:07.440 work with them, but there's a lot that will not. And it's a very difficult problem. And I can't help

01:52:14.240 because, I mean, I'm not an oncologist and I'm not qualified to be prescribing anything for people.

01:52:21.220 When people call me, I say, I'm happy to work with an oncologist. If your oncologist is willing

01:52:25.560 to try something different, I'm happy to at least give them my best sense for what the underlying

01:52:32.000 eco-evolutionary dynamics are. Understanding, of course, that a lot of times we don't really know

01:52:36.500 for sure, but this, again, in desperate situations, if they want to try this, that's between them and

01:52:42.020 their doctor. And as you know, those physicians are very aware of medical legal issues, are very afraid

01:52:49.680 of doing anything that's non-standard because it does open them up to lawsuits and that sort of

01:52:55.900 thing. And I understand that. And I'm not, that's not a criticism. It's just a statement.

01:53:01.100 So unfortunately, it's a very difficult problem. And it's a lot easier to deal with patients that

01:53:07.000 are just presenting with cancer than the ones who have been through many therapies, unfortunately.

01:53:11.820 You know, my father died of esophageal cancer. And I remember this vividly, the desperation and the

01:53:20.060 despair and everything that comes along with people at the end stages of the disease. And I would love

01:53:26.840 to help. I mean, I really wish I could, but unfortunately, most of the time, there's just

01:53:32.180 nothing left to do.

01:53:33.280 Well, Bob, I have found your, your story really interesting and really provocative. And I think

01:53:41.980 the frustrating thing about what you say is that there is, in my opinion, very little downside to

01:53:49.000 trying it. It would be very easy to conduct clinical trials that would have a crossover arm

01:53:54.140 where you could take this adoptive approach, pit it against a standard approach. And if the patients

01:54:00.840 on the adoptive approach were progressing, you could cross them over. And by the way,

01:54:05.860 vice versa, when the patients who are progressing on the standard therapy too quickly, you would

01:54:10.900 cross them over. In other words, when you think about some of the risks that we talk about in

01:54:15.320 medicine, this actually strikes me as a relatively low risk proposition. The greatest risk is to our

01:54:22.440 egos and to the dogma, as opposed to the patients. And that's, to me, a bit of the unfortunate

01:54:28.880 part of medicine is when we, and we're all guilty of it. I'm, I can probably count a hundred examples

01:54:34.640 of where my ego gets in the way, but, but this is a particularly sad example of it. When there is

01:54:40.520 a disease that we've really had very limited success treating, right? You referenced Nixon

01:54:45.280 earlier, right? The war on cancer is now in its 50th year and we don't have a whole heck of a lot to

01:54:54.320 show for it. Certainly not given the time and the resources that have gone into it. I don't,

01:54:59.620 I don't think anybody in that era in the early 1970s, if they had a crystal ball and showed them

01:55:05.900 what we have today would say that's success. Yeah. I mean, we, we certainly made some progress,

01:55:11.840 but we have not made a lot of progress in the metastatic cancer population, which is still more or

01:55:18.980 less as fatal as it was 50 years ago. And, and of course, well before then, I think it's important

01:55:24.740 to try to rethink some of the things we do that just keep doing the, doing the same thing over

01:55:29.240 and over again and just keep, you know, looking for new drugs. It certainly, you know, new drugs,

01:55:34.460 no doubt is important, but, but I think we could do better with the old drugs. And I don't think

01:55:39.380 that we've been really that incentivized. I'm not sure we've really taken a lot of time to think this.

01:55:46.000 And again, this is perhaps a bit of the, what we started with that intuitively it would seem like

01:55:51.300 killing as many cancer cells as possible is the best approach to, for the patient. And in fact,

01:55:57.980 in a nonlinear system, intuition is often not wrong, but nevertheless, it seems like it's, it's

01:56:04.500 right. I mean, we just, you feel absolutely certain you're correct, even when your own eyes tell you

01:56:09.560 that it's different. And that's hard. I mean, it's hard for all of us. Yeah. I mean, I think, look,

01:56:13.720 I think that statement is a great place to end this, which is in nonlinear systems,

01:56:18.740 your intuition can be very misleading. And that's, that's true in life of which biology,

01:56:24.580 especially this corner of biology happens to be exceedingly nonlinear. And with that, Bob, I want

01:56:30.280 to, I want to thank you for your generosity and time and more, much more importantly, of course,

01:56:33.560 the work you've done, which I think is helping a lot of people. And you're right. I mean, it might not

01:56:38.060 put a dent in the world of cancer in the next decade. But as we alluded to earlier, sometimes

01:56:44.680 some of the most interesting things in, in medicine take decades, plural to take hold. So

01:56:50.500 hopefully there are a number of people who are going to be able to pick up this baton and run

01:56:55.440 with you and eventually even be able to carry it. Thank you. Thank you. It's been a pleasure.

01:57:00.340 Thank you for listening to this week's episode of the drive. If you're interested in diving deeper

01:57:05.460 into any topics we discuss, we've created a membership program that allows us to bring you

01:57:09.860 more in-depth exclusive content without relying on paid ads. It's our goal to ensure members get back

01:57:15.640 much more than the price of the subscription. Now to that end, membership benefits include a bunch of

01:57:20.980 things. One totally kick-ass comprehensive podcast show notes that detail every topic, paper, person,

01:57:27.080 thing we discuss on each episode. The word on the street is nobody's show notes rival these

01:57:31.980 monthly AMA episodes or ask me anything episodes, hearing these episodes completely access to our

01:57:38.640 private podcast feed that allows you to hear everything without having to listen to spiels like

01:57:43.540 this. The qualies, which are a super short podcast that we release every Tuesday through Friday,

01:57:49.420 highlighting the best questions, topics, and tactics discussed on previous episodes of the drive.

01:57:53.680 This is a great way to catch up on previous episodes without having to go back and necessarily

01:57:58.540 listen to everyone. Steep discounts on products that I believe in, but for which I'm not getting

01:58:04.100 paid to endorse, and a whole bunch of other benefits that we continue to trickle in as time goes on.

01:58:09.600 If you want to learn more and access these member-only benefits, you can head over to

01:58:13.000 peteratiamd.com forward slash subscribe. You can find me on Twitter, Instagram, and Facebook,

01:58:19.760 all with the ID peteratiamd. You can also leave us a review on Apple Podcasts or whatever podcast

01:58:26.540 player you listen on. This podcast is for general informational purposes only and does not constitute

01:58:32.120 the practice of medicine, nursing, or other professional healthcare services, including

01:58:36.660 the giving of medical advice. No doctor-patient relationship is formed. The use of this information

01:58:42.560 and the materials linked to this podcast is at the user's own risk. The content on this podcast

01:58:48.400 is not intended to be a substitute for professional medical advice, diagnosis, or treatment. Users

01:58:54.800 should not disregard or delay in obtaining medical advice from any medical condition they have,

01:59:01.120 and they should seek the assistance of their healthcare professionals for any such conditions.

01:59:06.580 Finally, I take conflicts of interest very seriously. For all of my disclosures and the companies I invest

01:59:11.820 in or advise, please visit peteratiamd.com forward slash about where I keep an up-to-date and active list of such

01:59:20.900 companies.

01:59:41.820 Peteratiamd.com forward slash about where I keep an up-to-date and active list of such companies.

The Peter Attia Drive - October 25, 2021

#181 - Robert Gatenby, M.D.： Viewing cancer through an evolutionary lens and why this offers a radically different approach to treatment

Episode Stats

Length

Words per Minute

Word Count

Sentence Count

Misogynist Sentences

Hate Speech Sentences

Summary

Transcript