Watch Episode Here
Listen to Episode Here
Show Notes
This is a ~1 hour interview by Aastha Jain Simes (https://www.livelongerworld.com/) and I of leading cancer researcher Robert Gatenby (https://www.moffitt.org/inspiring-stories/dr.-gatenbys-story). Bob talks about the evolutionary and physiology approach to understanding cancer.
CHAPTERS:
(00:00) Evolutionary Hallmarks of Cancer
(12:02) Cooperation Stress Bioelectricity
(22:23) Information Thermodynamics Morphogenesis
(36:23) Cancer Information and Aging
(43:04) Adaptive Therapy and Biomarkers
(55:06) Habitat Imaging and Evolution
PRODUCED BY:
SOCIAL LINKS:
Podcast Website: https://thoughtforms-life.aipodcast.ing
YouTube: https://www.youtube.com/channel/UC3pVafx6EZqXVI2V_Efu2uw
Apple Podcasts: https://podcasts.apple.com/us/podcast/thoughtforms-life/id1805908099
Spotify: https://open.spotify.com/show/7JCmtoeH53neYyZeOZ6ym5
Twitter: https://x.com/drmichaellevin
Blog: https://thoughtforms.life
The Levin Lab: https://drmichaellevin.org
Transcript
This transcript is automatically generated; we strive for accuracy, but errors in wording or speaker identification may occur. Please verify key details when needed.
[00:00] Aastha Jain Simes: I'm really excited for this conversation because I think you bring such a multifaceted approach to looking at cancer from mathematical modeling, game theory, and an evolutionary lens. I would love it if you could discuss the new framing of the hallmarks of cancer, not as a static process but one which you see from an evolutionary lens. You've discussed how the tumor gets initiated, it changes its tumor microenvironment, and then you see this emergent phenomenon of cooperation among these tumor cells, which Mike talks about as well. How do you see this evolutionary progression of cancer?
[00:45] Robert Gatenby: From an evolutionary point of view, cancer is a fundamental evolutionary event. And whenever you talk about evolution, you talk about what's the unit of selection. And for our body, for the cells in our body, the unit of selection is us. Our cells have the same fitness as the three-dimensional multi-organ us. And they share that with us. They contribute to our fitness. And so they're controlled by the structures of our tissue, the local tissue instructions. Their death, their proliferation, their movement, their phenotype, all that is controlled by the tissue, by tissue signals. As a result of that, they do not evolve. Whereas in cancer, the unit of selection is the individual cell. So carcinogenesis means that there is a fundamental transition of cells that are defined by the fitness of their organism to cells that are individually subject to evolutionary selection. That's a tremendously important way to look at this. The traditional view of carcinogenesis is that it is accumulating mutations. In this setting, what we would say is that the cancer cells have to become independent of the host tissue signals. One of the problems with the sequential mutation model is the observation that you can see the same mutations in the apparently normal tissue adjacent to the cancer, which suggests it's part of it but not necessarily the cause of it. In the evolution model, we admit that you may get these accumulating mutations which renders the cell blind and deaf to the tissue signals. So they become totally independent because they can't receive them. On the other hand, an alternative is that you can damage the normal tissue. That damage prevents it from controlling the local cell population, which now allows them to evolve. And now the accumulated mutations that have occurred over a lifetime, which are similar all around, can become part of the fitness function of an individual cell. So there's a subtle difference in this. It's that transition to independence that's the critical function. And what this suggests is that cancer is not just a mutation. It's not happening to the cells. It's happening to the tissue. So damages to the tissue, inflammation, injury, or old age, where the tissues lose control of the local cell population, can then trigger this evolutionary process. When you think about cancer as an evolutionary process, the hallmarks of cancer need to be framed in terms of how they add to the fitness of the organism. One of the things that organisms do is called niche construction, beaver dam being the most obvious one. That's been added to evolutionary theory over the last decade or so as part of the inheritance process, because the beaver dam, as an example, can last for several generations.
[04:43] Robert Gatenby: It can persist as an ecological mechanism of inheritance. When you start to think in these terms, to me, a great example is the Warburg effect. This has been known since Warburg 100-something years ago, that cancer cells, even in the presence of oxygen, prefer glycolytic metabolism, meaning that they don't use oxygen to metabolize glucose, they just metabolize it to lactic acid. The problem with that is that it's very inefficient compared to aerobic glycolysis or aerobic metabolism. That's often called a dysregulation. There's something wrong with the cancer that it's not doing what it should be doing. Evolution doesn't do dysregulation. Evolution is the greatest optimizing force in the universe. If you're seeing the cancer cells use glycolytic metabolism in the presence of oxygen, that means it's conferring a fitness advantage. My own hypothesis is it's the acid that's produced that's the fitness advantage. It's a niche construction so that the cancer cells now can adapt to a high level of acidity, which normal cells cannot. It promotes invasion and it tends to blunt the immune response. It serves a purpose. That's the theme here: evolution is a scientific game of Jeopardy. Evolution gives you the answer. Our job is to provide the question. Evolution tells you that aerobic glycolysis is necessary. It increases fitness. Our job is then to find out why. It is not to say that evolution is making an error. That just will not happen. The idea that this aerobic glycolysis is a dysregulation, is an error in metabolism, and that somehow is at the root of cancer, is really missing the point. It's interesting to talk to pathologists who use stains to identify tumor cells, PSMA stain in prostate cancer, for example. You can then ask, why is the PSMA overexpressed? It becomes a circular answer. It's because prostate cancer cells make it, and that's how we identify them. No, it's not there for you. It's there because it's increasing the fitness of the cancer cell. Your job is to understand then why PSMA is increasing the fitness. It's nice to say that you can use that as a diagnostic tool, GATA-3 increased expression as a marker for breast cancer. But it's also evolution telling you that it's really important. That connection is often not made.
[08:44] Aastha Jain Simes: So instead of seeing the hallmarks of cancer as static, dysregulated traits, you see them as traits that evolve to enhance the fitness of the cancer cells in order to allow them to spread further.
[08:59] Robert Gatenby: This is evolution's message to us. Evolution is telling us this is what's important for cancer. This is what's necessary to optimize its fitness, which means to optimize its proliferation. We should listen to evolution more than we do. We talk about mutations. Clearly, if you see mutations in a lot of the cancer cells, then that mutation is optimizing fitness. But the opposite is also true. If you don't see mutations in the cancer cell in certain genes, those genes are necessary for fitness. Although we always talk about drugging mutations, and we have some math models that have looked at this, targeting the genes that are conserved is probably a more effective way to treat than the mutations, because those are unconditionally necessary for optimal cancer cell fitness.
[10:03] Aastha Jain Simes: Why do you say that? What does it mean to treat the genes that are not affected?
[10:08] Robert Gatenby: So what that tells you is if you assume that all the genes are undergoing random mutations, they then undergo what we've called evolutionary triage, meaning that if they increase fitness, the cell will proliferate. If they decrease fitness, the cell will not proliferate. So over time, what you'll see is that any mutation in that gene is selected out, meaning that evolution is telling you that a mutation in that gene decreases fitness. Evolution does not want that. That gene has to work perfectly. And so attacking that gene is attacking something that's very important to the cancer.
[10:59] Aastha Jain Simes: Does this feed into some of the atavistic theory of cancer as well, in terms of attacking some of the weaknesses of cancer instead of the strengths of cancer?
[11:11] Robert Gatenby: I think it does. The atavistic theory is that cancer becomes more primitive. There's something slightly pejorative about that way of looking at it, "oh, they're primitive cells." In fact, they've evolved to a very high state of fitness. So I admire them. They're deadly but beautiful evolutionary processes. They can basically withstand almost anything we can throw at them. So atavistics feels like we're minimizing the complexity of cancers.
[12:02] Aastha Jain Simes: Taking this fitness function forward, is it fair to view that, say, in normal cells, there's a degree of cooperation for them to function? And the cancer cell becomes an individual cell and adapts these functions to gain cooperation again, but it's a different level of cooperation at this point.
[12:29] Robert Gatenby: The word cooperation is so anthropomorphic that, I think you have to be careful about using that. The evolutionary biologists tend to use the word mutualism or something like herd effects. For example, herd effects being the most obvious one where these animals individually are the unit of selection, but groups of animals can act in a way that protects them all. Groups of cancer cells need to have at least a loose affiliation to work. For example, cancer cells need blood vessels, but it's probable that one cancer cell cannot produce enough angiogenic molecules to bring those in. The whole group has to do it. There's cheating that can occur. A cell in this group that's getting blood vessels in can decide not to make VEGF, for example. Now it's getting the benefit of the group but not paying its due cost. These can be very complicated dynamics, but the word cooperation implies sentience, that there's some thinking that's going on. I tend to like to avoid that. One of the things that I've noticed in working with oncologists is that there's almost a tendency to think magically about the cancer, that somehow it's plotting against you and has a remarkable capacity to overcome your therapy. They said there must be some inherent evil in it. I think it's more mundane than that, but it's not plain. They're not sitting down there, "gee, how can I get around this?" But they are extraordinarily powerful because they have the whole human genome to bring to bear on whatever evolutionary problem they face.
[14:53] Aastha Jain Simes: Mike, you often talk about cancer as dissociating from the collective behavior of cells. So what are your thoughts about herd behavior and evolutionary processes?
[15:08] Michael Levin: I have a couple of things to talk about. I think one thing I'd love to dig into a little more is the notion of stress initiating these processes. You mentioned tissue level stress. One of my favorite examples from our developmental biology world is this old, old finding that when you take a tail of a newt and you graft it to the flank, the thing turns into a limb. If you look at the cells at the very end of the tail, they're tail tip cells sitting at the end of a tail. There's nothing locally wrong. Nobody's poisoned, nobody's damaged. Yet they completely remodel. The issue isn't at the cell level. It's at a much higher level, but then filters down and makes the molecular biology dance to this higher level of error. My question is, what typically do you think the cells are measuring that leads them to be unhappy enough to break the multicellularity contract and say, this isn't working out for me. I'm going on my own. What's the trigger? Metabolic? What usually triggers this stuff? Do we know?
[16:20] Robert Gatenby: It's a great question because I'm not really sure. We've talked about this a little bit. There's a tendency to think about diffusing hormones and signaling molecules and receptors and things; that seems to be nowhere near enough to get the level of complexity that you see in cells, in tissue. It's not; it's their location, it's their phenotype. There's extraordinary synchronization of all this. I tend to think that, Mike, we've talked about this, we're missing a key information dynamic in cells. That is electricity, ions; there's communication that goes on that's not genetic. That is fundamentally important. When the human genome was being deciphered, there was a contest. This was around the world. Scientists from all over the world bet in a pool. It was called the gene pool. They each put in about $20 and bet how many genes would be in the human genome. Four or five hundred scientists across the world took part, and basically none were right. The mean estimate was about 60,000 to 70,000 genes. The upper ranges were in the 300,000s. The lowest was 27,000, and the number is under 20,000. David Baltimore wrote an editorial, published in Nature, and said, "We need to understand this. If we believe that we are the most complex of the organisms on Earth, why is our genome smaller than that of a mouse?" I think the argument I would have is that it's because the genome only indirectly controls the information dynamics. It sets up the dynamics that are based on ion conduction and on electricity flow that really are at the basis of complexity and of these interactions among cells. One thing that is interesting is that our cells have tremendous transmembrane ion gradients.
[19:21] Robert Gatenby: They're enormous. Mammalian cells spend about 1/3 of their energy budget maintaining these, which is about as much as they use maintaining the genome. I think in evolution you follow the energy. What evolution is telling you is that this transmembrane ion gradient is really important. It's so important that it devotes a tremendous fraction of its energy budget. And yet we know no reason for that. It serves no apparent purpose in the current understanding of biology. I think that's a tremendous gradient of information. Cells interact with each other, opening ion pores, which allows flux of ions across these gradients. Those fluxes are information. Another interesting factoid is that elements of the cytoskeleton are tremendous ion conductors. They're described as living wires. Why is that the case? There should be a purpose to that. Evolution doesn't do things with no purpose. I think that is a kind of wiring that allows ion flows across the cell to communicate within the cell. Mike does a lot of work with how cells communicate among themselves, all of which is through ions and conductivity. There's no genetics in that. The genes set up the dynamics: they make the pumps, they make the pores, and this is necessary for this other information dynamic, but I think it's an information dynamic that has been missed over the years, or at least hinted at, not necessarily in the mainstream of biology. To understand multicellularity, multisense, and complexity, I think we need to invoke a different information dynamic, going beyond the genome.
[22:23] Michael Levin: Do you want to talk a little bit about your recent paper on the computations in the networks within cells? I thought it was incredibly interesting.
[22:36] Robert Gatenby: Well, it really goes to that idea: there's something called Maxwell's demon. If you're familiar with that, Maxwell proposed this thought experiment. This was in the mid to late 19th century, when thermodynamics and Boltzmann kinetics were being discovered simultaneously. He said suppose you have two boxes with gases at the same temperature, and there's a hole between them with a gate, and a demon sits there. At any given temperature, you have this Boltzmann distribution of velocities. The demon would open the gate if a high-velocity molecule was coming toward it and close it otherwise. So essentially what he did was to shift all of the high-velocity molecules to one box and the low-velocity molecules to the other. This one box would be hotter than the other. It would look like a spontaneous flow of heat in isothermal conditions between two systems of the same temperature. It's also a spontaneous reduction in entropy. He viewed this as a violation of the second law of thermodynamics. It's been controversial ever since. But what it gave rise to in the early 20th century is information theory, because they said the demon has information. And so that's become a link. The ion gradient across the cell membrane is similar to what the demon does. What was really never appreciated is that once the demon has set up this gradient, you can have Maxwell's angels that can sit there looking out for information coming from the environment and open a gate. Ions can flow through this gate according to the gradient, and that is a flux of information. And so that's an important part of how cells interrogate their environment. The gross part of this is there's this molecular level difference. These are the ion gradients, but they converge to make a transmembrane potential. And this is at the border of ions. For Maxwell's demon, it was really the gas molecules changing the Boltzmann kinetics of each of the chambers, manifesting as a thermodynamic change. That intersection of microstates and macrostates is where information works. In this case, the information from the genome pumps the ions against the gradient using energy. I think you can calculate the amount of information within that by Shannon information criteria. It's enormous. It's orders of magnitude greater than the information content of the genome.
[26:51] Michael Levin: Another interesting thing you mentioned before about aging. We have this new work where we look at morphogenesis as a homeostatic process, which tries to minimize error from a certain set point. We did this simulation where we looked at what happens to this goal-seeking system in the cybernetic sense. What happens to that system after it has completed its goal? What does it do after that? What we found is that in the absence of any additional noise, thermodynamic damage, no entropic damage at all, the morphogenetic system reaches its goal and builds a nice pattern. This is my scheme that I've been developing: the goals are what keep collective intelligences together, and that after that goal is no longer active, they just start to go their separate ways in the absence of any external damage. It's fundamentally a cognitive problem, not a thermodynamic problem. How do immortal animals deal with this? Planaria, who are both extremely cancer resistant and ageless and highly regenerative, tear themselves in half every two weeks. They refresh this thing. They don't have a chance to settle down and wonder what to do next because they have this radical damage event that they cause themselves every couple of weeks. That seems to be enough to keep them rolling. I just wonder about the connection between aging and cancer. What do you think a stressor can be? We all know people who retire and don't know what to do, and then things go poorly after that. So it seems like that's a whole other new kind of stress. It's not that there's anything happening to you. It's that the goal setting is gone.
[29:40] Robert Gatenby: Maybe that's where exercise is creating these micro stresses, keeps us alert, keeps your body, keeps your cells in the game. But one of the interesting things that I've been thinking about is that the information in a gene is translated into a string of amino acids that can do precisely nothing until it folds. But that folding process is almost a probability matrix. It can fold into a number of different configurations. And because of that, there's actually information gain in the folding process. I'm not sure we've understood that. What's interesting then is that proteins that form enzymes, for example, are fine-graining because they act at a quantum level; the quantum interactions of the substrate need to be optimized to make the enzyme work, and fine-graining requires information. It's occurred to me that the way life does this is the protein folding: going from one dimension to three dimensions in a predictable way provides an information gain. I wonder if some multicellular organisms have that same process, where the cells are essentially point sources. As it proliferates, the interactions among the different cells are sufficiently predictable. Somehow in this fertilized egg, the information content is then magnified, increased by these interactions, these non-random distributions of the cells, so that the collective information of a multi-organism, which is vastly greater than that of a single cell, is gained by these dimensionality changes. It's hard to explain that, but biological information is not as simple as "the biological information is in the genome." How do you go from genetic information to the thermodynamic state of the cell, which is a highly, highly ordered, low-entropy, but far-from-equilibrium state? It's extraordinary. There's nothing in the universe like that. Intuitively, it seems clear that genetic information does that. But to me, it's not clear how you go from genetic information, quantified by bits, which has no physical unit, to thermodynamics, which is quantified by the typical units of thermodynamics; you don't have unit matching in this.
[33:08] Michael Levin: There is some theory around what bits — the minimal cost of a bit, the Bennett stuff and Landauer. But I think the bigger issue with the genome is that it's information about the hardware. And as you pointed out, it's some very cool reprogrammable hardware that it encodes. And after that, there's only so much you can blame on the hardware itself. There's a lot of physiological dynamics afterwards that are the main show, I think.
[33:42] Robert Gatenby: I've become interested in the idea of dimensionality, the linear string of the DNA folding into a three-dimensional configuration, a single sort of point source, a fertilized egg turning into a large three-dimensional structure. Those rules may be similar, I think, and the changing dimensionality may be how life exploits the limited information that comes from the genome to generate these large and complex organisms.
[34:42] Michael Levin: And that's just in 3D space. So you also get structures in physiological state space and transcriptional state space, not only metabolic; there are all these different spaces that it's unfolding into.
[34:56] Robert Gatenby: And then, to get really wild, you could say, well, what is the human mind, which is also unique in nature? You could argue that what evolution is going to select for are things that allow you to live longer, to survive, or to proliferate. There's no evolutionary selection to figure out the stars or to engage in quantum mechanics. It exceeds what evolution would have selected. My armchair speculation, which I'll deny ever saying, is that we recognize an additional dimension, time. Our ability to work in the dimension of time adds a dimensionality change which allows us to access information that would not ordinarily be present in strictly three-dimensional structures. I'll deny I ever said that.
[36:23] Aastha Jain Simes: Going back to information dynamics, how do you think about information dynamics for a cancer cell compared to a healthy cell?
[36:33] Robert Gatenby: I think there are two components to this. The healthy cells are really tuned into their environment and to the other cells in the environment. They're tuned into this collective; I think Mike used the term. Cancer cells have to be independent of these control mechanisms. At the same time, cancer cells have to be able to forage. They have to be able to detect sources of substrate, and to detect bad environments, and ideally detect predators like the immune system. So the information dynamics change. That's where I think calling them atavistic is insulting to them. They are extremely fine-tuned to optimize their proliferation in the middle of your body. And that's often not a very good place to be. Some of them grow in the lymph nodes, robbing from a police station. The whole immune system is around them, and yet they continue to grow. So they need sophisticated information dynamics, but these are different from what normal mammalian cells have. What I think is really important to cancer is that in normal mammalian cells a lot of the genome is blocked. They don't have access to a lot of the genome. Cancer seems to be able to access the whole human genome. So they find genes that were being expressed in the fetus but weren't expressed in adults, and they can use that. They find ways to evolve and to optimize their fitness using parts of the genome that ordinarily are not accessible. I think it's remarkably clever and really scary because they're really good at it.
[39:01] Aastha Jain Simes: But I suppose from an evolutionary perspective, evolution would want us to reproduce and live as long and pass on our genes. So the information that it would pass to us would be the ones that normal healthy cells have. Where do you think cancer cells are acquiring this information per se? Or do you say maybe it's the fact that evolution has just programmed us for fitness and cancer cells just have this ability to figure out whatever is needed in order to proliferate.
[39:35] Robert Gatenby: I think, after you've reproduced, and I'm sorry, I know you have a new baby at home, but after you've reproduced, nature would rather you were dead. There's the whole grandmother hypothesis and various things that increase fitness of offspring. But if you're a man at my age, evolution would rather I just drop dead because that's why I exercise — I'm trying to convince my body that I'm still part of the hunt and it'll let me live. I think that's a really important point. There's no evolutionary selection for anti-cancer mechanisms beyond the sort of reproductive age. We live far longer than our reproductive age, unlike most animals. So there's unfortunately no selection for people who get cancer in their 60s and 70s. It's just because we've reproduced. So evolution is pretty harsh.
[40:59] Aastha Jain Simes: But if you think about it from an information standpoint, because cancer cells also have some information embedded in them in terms of how they want to behave, do you think after we've reproduced the information dynamics change in the sense that the information that the evolution program does for being healthy shifts or no longer cares about that? What I'm trying to understand is how cancer cells acquire this different set of information dynamics.
[41:29] Robert Gatenby: I think that James DeGregory is a great scientist in Colorado who looks at aging. One of the things that happens with aging is that there seems to be a loosening of the tissue controls over the individual cells. That probably contributes to the fact that it's a process that occurs generally in older adults. I don't think the information cancer uses was necessarily there for the cancer's benefit. It was there because it benefited fetal cells or there's some part of it that the cancer cell can then adapt. It's not just the gene itself, but there's the spliceosome. There are a number of things that the cancer cell can change, not to mention the role of the environment, acidosis, things like that. It can change the environment in ways that are beneficial. I will not pretend to know all of the details of this.
[43:04] Michael Levin: Do you have any — you must get this way more than I do — we publish some things on cancer and we have a lot of patients writing to us asking what to do. Aside from following whatever your best oncologist is telling you, do you have any other favorite thoughts for people to look into, whether that be Seyfried's metabolic stuff or anything else? Do you have anything that you tell people?
[43:36] Robert Gatenby: As a cancer prevention agency agent, is that?
[43:38] Michael Levin: I think we're beyond that. Once somebody's already been diagnosed, any suggestions there?
[43:45] Robert Gatenby: Cancer is an evolutionary process. Cancer, like all living systems, has to obey the laws of evolution. Cancer is not magic. It's not an evil entity. It's just an evolving population. We know the rules of evolution, and we can use those to optimize cancer therapy. For example, the standard practice in oncology, which has been the case for five decades, is that any therapy is given at maximum tolerated dose continuously until progression. It's almost a mantra. Evolutionarily, you could hardly do it worse because you're placing strong selection pressure for resistance on a large and diverse population. You're also removing their competitors. You're killing the cells that are sensitive. You're selecting the ones that are resistant; these resistant cells have no competitors, so they're free to proliferate. This is well known. This is a phenomenon that's observed in evolution. It's called competitive release. This is a strategy that optimizes the proliferation of resistant cells. One of the things that we've tried to do is to bring evolutionary dynamics into cancer therapy. For example, we've used something called adaptive therapy. Here, instead of giving maximum tolerated dose continuously, you give a reduced drug dose. In prostate cancer, for example, we use the serum biomarker PSA and just drop it 50%. Then you stop; you withdraw it. The tumor comes back, but the tumor is now proliferating with no selection for resistance. Because in general, in the absence of therapy, the resistant cells have some molecular machinery that allows them to be resistant, which gives them a fitness advantage.
[46:46] Robert Gatenby: When therapy is present. But in the absence of therapy, it's a cost with no benefit. And so in general, the sensitive cells are fitter than the resistant cells in the absence of treatment. The tumor comes back, and it recapitulates the distribution of sensitive and resistant cells that was there at the beginning, and you just treat it again. You just keep pushing it down. This is in metastatic cancers that are fatal. The goal is to maintain life with the highest quality possible for as long as you can. What we've learned is that, in this kind of approach, we're only giving about half the drug that you would ordinarily get. The side effects and the costs of the drug are much less. But we can prolong life considerably with that kind of approach. It doesn't have the sexiness of trying to kill all the kingpins. That intuitively is what we would like to do. That would be the right thing to do if you could. But given the fact that we don't have magic bullets, we don't have treatments that will kill all the cancer cells. A standard example is androgen deprivation therapy for metastatic prostate cancer. It's really effective. Ninety, ninety-five percent of the time, it reduces the PSA to normal or even unmeasurable, but it's also never curative. We know that historically it is never curative. That's the dynamic. And the goal is then to reduce the amount of ADT that you give, androgen deprivation therapy, which men hate by the way. It has a lot of toxicity, significantly decreases quality of life, and typically gives you maybe three years. But if we can do much better than that so that they have a normal testosterone level about half the time, we've both reduced the cost and we've reduced the toxicity while also increasing lifespan. So that's the approach that we've tried to take, and that's worked quite well. But it's so different from standard cancer therapy that it's really intuitively unappealing to oncologists and sometimes the patients, because you drop the PSA to 50% of its original pre-treatment value. There's something that says, "let's just keep hitting it," that that's the right thing to do. So you have to think through that and overcome that intuitive sense that you're not doing it right.
[49:47] Michael Levin: And is that thing just in terms of availability — you guys are doing it at Moffitt? Is that it or is this accessible to people? How is this looking out?
[49:58] Robert Gatenby: I think there are large trials going on in Europe and Australia. There's one using this approach in ovarian cancer that's going on in a multi-institutional study in the UK. There are some physicians and oncologists who are willing to do this, but it would still be considered non-standard practice. And so it will take a long time to change ideas about this.
[50:40] Aastha Jain Simes: If you were to hypothesize, where do you think adaptive therapy might not work the best?
[50:47] Robert Gatenby: We don't know. I don't think we understand immunotherapy sufficiently well. The basic rules should be the same. Evolution is evolution. But what we don't know is we have the cancer cell population, which can evolve, and we have the immune response, which can change. Those are very complex dynamics. We're just beginning to look at those dynamics because they're very complicated. There are so many things that are going on. It's difficult to put all that together. It's much easier to work with hormonal therapy and chemotherapy. Targeted therapy seems to be slightly different. But I will continue to argue that fundamentally evolution is a first principle. Evolutionary dynamics will hold. They simply — we don't know how they play out in some of these treatments. But I think you can argue when I first was working at a cancer center, which is a long time ago, there just weren't effective therapies for a lot of cancers: lung cancer, renal cancer, melanoma. There was nothing. What we have now is really pretty good therapies for almost every cancer. Yet metastatic lung cancer is fatal now as it was then. The reason is evolution: the cancer cells evolve resistance to an initially effective therapy. I think increasingly you could argue that evolution is the proximate cause of death in cancer patients. And I think unless we acknowledge that and unless we start to integrate evolutionary principles into our therapy, we're not going to make progress, unless we find a magic bullet, which may or may not exist.
[53:09] Aastha Jain Simes: What are some of the open questions you're tackling now by integrating more of the evolutionary dynamics into cancer?
[53:19] Robert Gatenby: One of the things we need desperately is biomarkers to understand intratumoral evolution over time. We don't have a good way to estimate that. Recently, sometimes you just need more data. A lot of our trials have been done on shoestring budgets, so we didn't necessarily have the ability to do some of the tests that we would like to do. But something as simple — what we found recently is, if you take the testosterone level and the PSA level simultaneously, the ratio of that can tell you a lot about the subpopulations. So it doesn't have to be esoteric circulating DNA. We can use standard tests by understanding the connection between them. What we find is the PSA should reflect the testosterone level in general. If, for a given testosterone level, the PSA starts to increase over time for that individual, that means there are cells present that are able to use much lower concentrations of testosterone to proliferate. They're making the PSA, but the testosterone level is low. That allows us to understand that a population of resistant cells is increasing and we need to do something about it.
[55:06] Aastha Jain Simes: Interesting. I know you've also discussed this idea of habitat imaging and looking at the tumor microenvironment itself and then personalizing protocols based on that. I'm curious how that personalization gets done. Why is this not being done right now?
[55:29] Robert Gatenby: Habitat maps, species maps. So if you look at Florida, for example, and say, what's the distribution of ground squirrels? They'll show you maps of whether the gray squirrel is dominant or the fox squirrel is dominant. People don't walk around every square meter of Florida measuring squirrels. And the way it's done is what's called landscape ecology. This was developed for satellite images and large spatial scale images. By identifying habitats, investigating a habitat and counting the squirrels in that one habitat, you can then do a distribution. So for example, gray squirrels are very good at foraging. They can out-compete the red squirrels and the fox squirrels. So if you see a college campus, you will always see gray squirrels there. What fox squirrels and red squirrels are good at is dealing with predators. They'll go up a tree and attack an owl. So in forests and in wildlife areas, the red squirrel will dominate, the fox squirrel will dominate. You won't see gray squirrels as much. Now, on the other hand, if you see gray squirrels on a college campus, that means either there's coyotes there or there's feral cats. Evolution tells you the truth, and our job is to understand. In this case, when we do imaging with radiologic studies, these are large scale, the equivalent of satellite images. If we can identify certain habitats in which there will typically be certain kinds of tumor cells. So for example, if it's an area that's very poorly perfused, we would expect cells that are hypoxic, that are acidic, they have those kinds of capacities. By understanding how the large scale environment selects for the small scale population, you can then estimate the subpopulations of the cancer. The problem is that it's really hard to do that. What we did was take MRI scans. In MRI scans the same tissue is repeatedly interrogated with different sequences that are sensitive to different components of the tissue. Putting those together, we can generate a habitat. But this is technically very difficult because the sequences are never precisely the same. The spatial scales are often slightly different. For example, there can be a three-millimeter slice thickness versus five millimeter, seven millimeters. Technically, there's a lot of things to overcome. What we have pleaded with the imaging companies to do is to develop a sequence that allows you to interrogate the same tissue with multiple sequences simultaneously, which in theory is possible. But that's been very slow to develop. It's been a frustrating path, and this is sufficiently expensive that we need grants for that, and they have not been forthcoming.
[59:37] Aastha Jain Simes: Mike, any other questions?
[59:40] Michael Levin: No, a lot to think about. I think this consilience of evolutionary kinds of considerations over the population and the sort of software aspects in the decision-making of the collective and these physiological circuits, I think this is where in the end the solutions are going to come from. I think we really need to understand both sides of it.
[1:00:09] Robert Gatenby: It's funny, if you go back to the era when the human genome was being deciphered, if you read the literature, it implies that this is pretty much it. This is the end of disease. I think that term was even used. It's funny because we still live in the genetic era where everything that's done is based on genetic measurements. I think that a whole generation has been almost lost because there's been so much focus on that we've given up on many of the other things which were actually being developed in the 1950s and 1960s but then got passed over by the genetic revolution. It's just so easy to put your sample in a molecular biology machine and generate data. And I think there's a false sense that all this data is going to tell us something really great. An interesting analogy is to say, suppose you were a modern Darwin and you were on the Beagle with just $1,000,000 worth of molecular biology machinery. Suppose the sailors tramped through the Galapagos Islands and they brought you back pinch samples, and you ground them up and put them into your molecular biology machines and generated terabytes of data. Could that modern Darwin have written "On the Origin of Species"? I think the answer is no. Although I've posed this to molecular biologists who always say we could have done it. But the problem is that there's not a clear mapping from genotype to phenotype. So you would have to know from the gene that there's something about the beak, but you cannot see the selection force, the morphology of the beak. What Darwin saw was something very straightforward and logical: the morphology of the beak and the morphology of the seed matched. For all the molecular data we generate, I don't know if we've gotten a true sense of what's really going on. I think we've missed the seeds. In cancer, at least, we've not done the phenotype, the beak versus seed kind of understanding. So we've got lots of genetic information, but it's not being built on a solid framework of evolutionary first principles.
[1:03:23] Aastha Jain Simes: Your framing of where evolution is spending energy and how one-third of it is even just ion channel transport tells you a lot about where we should be focusing our efforts.
[1:03:36] Robert Gatenby: Follow the money. I think evolution is trying to tell us something. We've not been listening very well up until now.
