Cancer as Evolution, part 3

Kevin points out that perhaps I am giving chemotherapy short shrift and not looking at the bigger picture.  I would counter that you can’t really make the latency analogy with human life because we are all born terminal.  “Mean survival time” is a squirrely measure at best because how do you know if someone died (a) because of the cancer, (b) because of the chemo, or (c) because of some other factor (in which the cancer or chemo or both) could be confounding?

If you buy the somatic evolution (SE) argument then there are all sorts of consequences which contraindicate chemo in most cases. Just for instance, there is good experimental evidence that SE is punctuated (as in “punctuated equilibrium”).  If you introduce stress into the population at the wrong time, you tip the system into a regime that is bad for the human (good for the cancer cells).  A more auspicious time to apply chemo (according to SE) might be before symptoms occur.  But doctors and patients are perhaps more likely to be aggressive with chemo at precisely the wrong time.

Another consequence of SE is the “cleverness” of the disease in terms of routing around the brute-force obstacles we throw at it.  Chemo is just the sort of unspecific and mild selective pressure that evolution is really good at adapting to.  So instead of mere drug resistance, chemo is really creating a new system which thrives in the presence of chemo; if you don’t kill every potentially bad cell, you could be worse off than when you started.  But comprehensiveness is hard to achieve because after all, the cancer cells are yours to begin with and there is strong selective pressure for them to learn to evade detection, not play nice in the cell-signaling game, etc.  The only way we know how to tell whether a cell is cancerous or not is by observing its behavior in the presence of other cells in the body of the organism in question.

I would like to re-iterate, I am not bashing chemo per se, just that it’s in a class of potential solutions which are based on a theory of what cancer is that is fundamentally incompatible with the evidence.  Furthermore, one theory which is compatible with the evidence — somatic evolution — suggests that if you start with a chemo approach, you will always be patching a leaky boat with material that is corrosive to the boat itself.

A final piece of circumstantial evidence comes from an experiment you can do yourself.  Find as many practicing oncologists as you like, get them to speak to you off the record, and ask them the following question: “If you or a family member was diagnosed with a common form of solid cancer (breast, colon, etc), would you administer chemo?”  Previous such surveys suggest that less than 25% would choose chemo for themselves, but nearly all of them would for their patients (due to malpractice concerns).

BTW, somatic evolution is not the end-all-be-all in cancer theory.  There are other dynamics at play too.  Or more precisely, evolutionary theory as currently understood by most people in life sciences is not complete enough to account for these other important dynamics.  One of these that I think needs to be reconciled and explored in much greater depth is what’s known as aneuploidy.  Aneuploidy as it applies to cancer refers to any sort of damage done at the level of the chromosome (as opposed to the gene-level, which sits below the chromosome level).  In cancerous systems, the gene-level dynamics can look extremely stochastic, while at the same time, genome-level (i.e. chromosome-level) dynamics show remarkable patterns.  Why this isn’t addressed in most cancer research at this time is a larger discussion about how science advances in the real world.

  • kevindick

    Let’s not say chemo therapy is bad/wrong/misguided because of somatic evolution. Let’s make some specific claims. Here are some alternatives:

    (1) As a society, we shouldn’t spend very much scarce research money on traditional chemotherapy approaches.

    (2) As an individual, I should forego chemotherapy if diagnosed with cancer.

    I think cancer as evolution is strong evidence for (1). I think you need to look at much more specific evidence when considering (2). For a given cancer at a given stage, you can literally look up the mean survival times with and without chemotherapy from a controlled trial. If the difference seems worth the expense + side effects, you take the chemo.

    Are you really claiming (2) as a general rule? Now, I could easily be persuaded that it should be the _default_ rule. But in the face of a well-designed, applicable study that says I’ll live significantly longer with few side effects?

  • Let me know when you find that well-designed, applicable study which says you’ll live significantly longer with few side effects….

  • kevindick

    Well, I randomly chose breast cancer and Googled “breast cancer chemotherapy survival time”. This was the first result:

    http://www.sciencedaily.com/releases/2006/12/061215123112.htm

    Now, if it were my wife, sister, or mother (assuming they were HER-2 positive), I’d recommend they get the Herceptin–after getting a copy of the original study, verifying this account, making sure this holds up when compared in other studies, and getting a full cardiac work up.

    I could have been incredibly lucky and found the one decent chemotherapy treatment in my first try, but I doubt it. My inference is that there exists some set of currently available chemotherapies that have a benefit-cost ratio greater than 1.

    Did I miss something?

  • You’re missing the fourth arm in the study (which wasn’t done): active placebo. And after reading the report three times, I still can’t tell whether they tried Herceptin only (no chemo).

    Herceptin is not intended as a cytotoxin while “chemo” as I am using the term is. The point being that killing cells in an indiscriminate way provides selective pressure that can promote cancer (if you don’t kill all “bad” cells before you kill the patient). Killing cells in a targeted manner does work better (surgical excision being the prime example). Interacting in a sophisticated and targeted way with the biodynamics of the system as a whole is even better and jives better with somatic evolution.

    The problem in all of this is that it’s very hard to get specificity (good targeting) and comprehensiveness. For a very sophisticated approach to this, I recommend Arny Glazier’s work.

  • kevindick

    Ahh, I see the distinction. I was thinking of chemo as “current drugs”. You’re saying that traditional chemo is defined as broad spectrum cytotoxins.

    So targeting cancer cells, either through a specific biological mechanism of action or through some physical delivery system, is not categorized as traditional chemo.

    It seems then that we should be more sophisticated in our analysis of alternatives, perhaps rating interventions on how “targeted” they are? So a surgical excision with clean boundaries is highly targeted. Herceptin is somewhat targeted. Andriamycin is almost non-targeted.

    I will update my Bayesian priors on cancer treatment effectiveness based on this “targetedness” metric.

    This helps explains the pre-chemo starvation approach. You’re attempting to enlist the body’s own metabolic machinery to make otherwise broad spectrum compounds target cancer cells specifically.

    Good stuff! I feel I have gained my important insight for the day ;-)

  • Hi Rafe –

    I am a scientist who also happens to be a poker player. I come from an academic background, but most recently have been working in the Boston biotech/pharma area, most recently in the discovery dept of an oncology based company.

    I want to point a few things:

    1) somatic evolution: this looks like a garbage idea. The idea of evolution without giving rise to self-sufficient organisms is silly. Scientists have known for a long time that the somatic cells that make up our body have limited reproduction potential, and with the limit comes errors as they approach the end of their reproducability. These errors are typically what gives rise to ‘rogue’ cell colonies which draw to heavily on shared resources, ie cancer. To think of this as evolution is a complete twisting of the concepts underlying evolution.

    2) Nearly all chemotherapies are cytotoxic. You have to kill the cancerous cells, you hope not to kill the healthy cells. Think of it as replacing the surgeon’s blade with the chemist’s molecule. Some chemotherapies are well tuned (a result of ongoing research and development) and leave the healthy cells unharmed, others are blunt instruments. You start with the well tuned ones and proceed to the blunt ones depending on patient response.

    3) Life extension: I’d say the importance of this varies from person to person. In any case, 3-6 months life extension is generally considered successful. Enabling life to the point where one dies of something other than the cancer being treated is considered a cure. If 3-6 months sounds short, consider this question: Do you want to die in two weeks or do you want to die in 3 months? How long does 3 months look now?

    4) The part of cancer where we have made little progress is in metastasis, which is the spread from the initial ‘rogue’ colony to other parts of the body. Focused treatments are available for most types of cancer (skin, lung, gi, etc) with varying amounts of success. Once a cancer decides to visit other parts of the body, all bets from modern medicine are off and we are back to only slightly better than medieval practice levels, with corresponding results.

  • Carlo Maley

    Biggestron, in reference to your point 1), I agree that it seems counter-intuitive that cells in a multicellular body could evolve, but evolutionary biologists have known for a long time that they actually do. This is known as multilevel selection in the field and can be seen at a variety of levels of organization (genes, cells, individuals, etc.).

    In a multicellular body, eons of selection at the individual level have led to the evolution of genes (like tumor suppressor genes) that enforce cooperation between cells. However, natural selection happens any time that there is heritable variation in the population (of cells, in this case) and that variation affects survival and/or reproduction (of cells). Because the process of copying the genome is not perfect, errors are made at each cell division (though at low frequency). This happens even before old age, when cells may have reduced their telomeres to the point that they can become increasingly genetically unstable. Environmental exposures like smoking or UV light can increase the somatic mutation rate. That leads to heritable variation among somatic cells. Occasionally, those errors inactivate a tumor suppressor gene (like p16 or p53), or activate an oncogene, and give the cell and its progeny a survival or reproductive advantage. If you are in oncology research, then you are probably aware of the clonal expansions generated by clones with the competitive advantage over other somatic cells (for references see Wikipedia’s page on Somatic Evolution in Cancer http://en.wikipedia.org/wiki/Somatic_evolution_in_cancer). You are also probably familiar with Hanahan and Weinberg’s hallmarks of cancer. All of those hallmarks are phenotypes that give a clone a competitive advantage over other clones that lack the hallmark.

    All of the above observations do not imply that cells are adapted to be cancers. In fact, they start the process well adapted for being parts of multicellular bodies. The somatic evolution that occurs appears to mainly destroy the mechanisms that enforce that cooperation. You are right that the tumor cells are not self-sufficient organisms, but that has never been a precondition on evolution. All viruses and many single cellular and even multicellular organisms are obligate parasites. What is unusual about cancers is that they have a relatively short evolutionary history – arising anew in each host. However, the genetics of that are probably ancient. The evolutionary transition from single cellular life to multicellular life can be viewed as the evolution of mechanisms to enforce somatic cell cooperation – in other words, to suppress cancer. Cancers can then be thought of as the failure of those systems and the reversion to a single cellular (obligate parasite) organism. The process by which that happens is a process of somatic evolution. This has important implications for oncology.

  • Biggestron,

    Thanks for taking the time and interest to read and comment here. Carlo did a better job of making the case for somatic evolution than I could, but I will address your other comments.

    2) For those (like myself) who are convinced that somatic evolution is the reason cancer exists, there are some important implications for chemotherapies, targeted or not. One is that you can predict drug resistance similar to that in fighting infectious diseases (which is after all fighting an evolving population). The second is that targeting really becomes key. Currently we have very blunt blades, as evidenced by the nearly flat age-adjusted mortality rates since the 1950s (while during the same period heart disease mortality for instance has dropped significantly). Thus, if we want to give people more than 3-6 months (of poor health) we had better get a lot better at targeting. Unfortunately, somatic evolution makes it really hard since the target is always moving, like a school of fish that largely evades its predators despite the fact that a few of them are successfully targeted and eaten.

    3) If you were pushing a bolder uphill to roll it down the other side and I told you that it will get exponentially harder the higher you go and that you will never get to the top, would you still like your approach? Would you perhaps look and see if it’s possible to roll the bolder around the mountain instead?

    4) Yes, you are right about this. And once again, somatic evolution has implications here. Metastasis is in fact the school of fish avoiding its prey, which includes the immune system as well as chemotherapy. What we should be careful about is how and when we apply chemo, because it is currently so blunt, even the targeted ones. And the less targeted the chemo, the more selective pressure is provided for the colony of rogue cells to evolve resistance.

  • It was more the name ‘somatic evolution’ that gave me problems. I have no doubt that cells change as they reproduce, and that some of those changes may result in a different microenvironment for those cells and their neighbors (ie colonies). That was why I brought up the idea of ‘limited reproduction potential’. Errors during DNA replication and chromatin repacking (chromatin are the larger particles comprised mainly of DNA and other proteins) are what gives rise to changes. In the case of single cell organisms, you can call this evolution. For multi-cell organisms, you can really only call it evolution if the germ line (eg sperm/egg) is affected. The great idea behind evolution is that the changes do *not* occur in response to the local environment, but rather that they are completely random and confer some advantage to the environment. For example – it is a mistake to say giraffes grew long necks to reach the leaves in trees. Rather, giraffes with longer necks arose as a natural variation and could reach more food giving them more resources and therefore better reproduction potential.

    Perhaps all of this is a little bit of nit-picking, but you invite it when you misuse a well accepted term such as evolution.

    The issue of viruses and obligate parasites are at the blurry edges of what defines ‘life’, so I won’t argue those points.

    I agree that changes in somatic cells are behind much (if not all) of cancer, and that stopping such changes are indeed a promising route to beating the disease. It is fairly difficult to stop DNA errors – organisms are “designed” (perhaps better to say ‘have evolved’) to make the occasional mistake in their DNA so that evolution and thereby adaptation can occur for the species. However – remodeling of the chromatin (through histone modification) is an active area of oncology research. Look into the field of epigenetics to see what is going on there (http://en.wikipedia.org/wiki/Epigenetics) – it is a potential explanation for the disconnect between the presence of an aberrant gene and the seeing or not seeing an end result of phenotype/disease. There are a few start-ups in the Boston area pursuing these ideas for new oncology treatments (Constellation http://www.constellationpharma.com/epi_newscience.htm, Epizyme – google them for more).

    And btw – one more nit to pick: metastasis is not ‘in fact’ a ‘school of fish’. That is a good metaphor, but it is not a fact. There are some who believe that metastasis is the result of only a few mobile cancerous cells who are able to invade other tissue colonies (eg other organs) and ‘fake it until they make it’ – ie convince the other tissues to accept them and start using up resources. Ideas behind stopping metastasis are probably the ones in shortest supply – some of them center around the ‘stem cell’ idea of oncology (again, these aren’t actually stem cells per se but the buzzword has been stolen to describe them).

    I’d best stop here before I head off on even more tangents…

    In any case – I don’t mean to be critical in a negative way. I am all for non-scientists (or rather non-professional scientists…) making an effort to understand what is an extremely difficult and contentious topic. You are doing a great job here!

  • @ Biggestron:

    “In the case of single cell organisms, you can call this evolution. For multi-cell organisms, you can really only call it evolution if the germ line (eg sperm/egg) is affected.”

    With all due respect, I think you are missing the crux of what evolution really is. I recommend you re-read Carlo’s comment here: “However, natural selection happens any time that there is heritable variation in the population (of cells, in this case) and that variation affects survival and/or reproduction (of cells). Because the process of copying the genome is not perfect, errors are made at each cell division (though at low frequency).” For more depth on this, see my post on generalized evolutionary theory. For further depth, including rigorous mathematical analysis of the concepts which Carlo and I refer to informally, check out Harvard professor Martin Nowak’s recent book Evolutionary Dynamics (or even John Holland’s seminal Adaptation in Natural and Artificial Systems). For even more current and focused discussion of cancer evolution, check out David Basanta’s cancerevo blog.

    “The great idea behind evolution is that the changes do *not* occur in response to the local environment, but rather that they are completely random and confer some advantage to the environment.”

    This a red herring. While it is true that randomness to a certain degree is necessary to generate novelty and diversity from which differential fitness (and thus selection) occurs, there is quite a lot of debate over the “completely random” claim. Check out facilitated variation, which if you read their book is a generalization and explication of the the Baldwin effect.

    The overall point I would like to make here is that while Darwinian evolution of the human species plays a role in the overall cancer equation, a much more direct impact (what I would even call a defining characteristic) is somatic evolution, that is, the Darwinian evolution of the population of cells within a single human body during their lifetime. These two dynamics — human evolution and somatic evolution — are NOT the same thing. They are based on the same underlying principles, but they occur on very different timescales, orders of magnitude different. I am not claiming, for instance (as I think you may have implied), that cellular changes within the soma have an impact on human evolution. They do have an impact on somatic evolution, which is, according to many people, what cancer is all about.

  • Dear all,

    Seems to me that Carlo and Rafe had already said what I would have said if I had had the chance of posting this faster. In any case here goes my take on why thinking of cancer as an evolutionary disease is not garbage.
    I’d first point out that I am not sure what Biggestron means with evolution not making sense without self sufficient organisms. My point would be that, depending on your take on it, there are really no self sufficient organisms in the sense that we all require some environment and resources in which to live. Same with cell colonies. Same with a tumour in which cells have acquired some degree of self sufficiency with the environment (which is one of the hallmarks of cancer progression as stated in Hanahan and Weinberg’s widely cited paper in Cell in 2000).

    In any case what you need for evolution is exactly three ingredients, no more and no less: variation, heredity and selection. This is not my definition but something that can be found in many references dealing with both natural and artificial (eg. evolutionary algorithms) evolution. These three elements can be found in cancer as multiple clones arise and those clones compete for the limited resources (mainly space and nutrients) available. This has been known and widely accepted at least since Nowell’s paper in 1976 (P.C. Nowell, Science, 1976). Asides from the references mentioned by Rafe I’d recommend Merlo’s paper in Nature Reviews Cancer in 2006 (I think in Maley’s group), Crespi and Summer’s paper in Trends in Ecology and Evolution in 2005, Moffitt’s Smalley (Seminars in Cancer Biology 15 (2005) 451-459) and Gatenby (Nature Reviews Cancer, 2008 vol. 8 (1) pp. 56-61) or Anderson (Nature Reviews Cancer, 2008 vol. 8 (3) pp. 227-234).

  • Cancer cells are once healthy cells that have mutated in a way that confers a growth advantage at the expense of the entire organism. Because the replication of individual cells necessarily involves duplicating the DNA, we will see the same mutated DNA passed on to future generations of cells. If these cells grow faster because of the mutation, we will see a colony emerge. In some cases, phenotypes may arise not from the classic base pair (A T C or G) type of mutation but rather from epigenetic changes that are inherited by daughter cells. There is much to be learned of this latter aspect of chromosomal inheritance.

    However – calling this effect ‘evolution’ for somatic cells is a stretch. It seems like you are considering our body as being composed of multiple obligate symbiotic organisms rather than individual cells, tissues and organs. We can carry it further and start calling normal development ‘programmed multiway co-evolution’. Like the term ‘somatic evolution’, It sounds cool, but it adds nothing to our understanding.

    You can call the concepts of modern oncology by whatever buzzword you wish. You can describe the orbital motions of planets in our solar system with the earth as a centre, but it ends up making everything unnecessarily complex

    In short, kudos to the people who work on cancer cures and to those who would make the attempts to understand that work. However – I don’t like the term ‘somatic evolution’.

    Just my opinion.

  • @ Biggestron

    Cancer cells are once healthy cells that have mutated in a way that confers a growth advantage at the expense of the entire organism.

    More specifically the mutations confer a growth advantage at the expense of normal cells. This, critically, is why it is evolution and not some evolution-like developmental change.

    Because the replication of individual cells necessarily involves duplicating the DNA, we will see the same mutated DNA passed on to future generations of cells. If these cells grow faster because of the mutation, we will see a colony emerge. In some cases, phenotypes may arise not from the classic base pair (A T C or G) type of mutation but rather from epigenetic changes that are inherited by daughter cells. There is much to be learned of this latter aspect of chromosomal inheritance.

    Yes, this is a good point. Evolution (whether you are talking about Darwin or the modern synthesis) has nothing to do per se with DNA. DNA just happens to be the phenotypic encoding mechanism that life on Earth settled upon. You can describe evolution without ever distinguishing between genotype and phenotype and it works just the same. So, these epigenetic changes are indeed worthy of study and they potentially help solve some mysteries of cancer, like the high amount of stochasticity and lack of discernible pattern at the gene-level.

    However – calling this effect ‘evolution’ for somatic cells is a stretch. It seems like you are considering our body as being composed of multiple obligate symbiotic organisms rather than individual cells, tissues and organs. We can carry it further and start calling normal development ‘programmed multiway co-evolution’. Like the term ‘somatic evolution’, It sounds cool, but it adds nothing to our understanding.

    On the contrary, it adds to our understanding by allowing us to take the lessons learned from organismal evolution and testing whether they add understanding to cancer formation and progression. One such glaring insight is the area of drug resistance. If cancer is somatic evolution then drug resistance is a likely consequence, similar to what we find in evolving populations of infectious disease. Indeed, the epidemiology of cancer matches this hypothesis perfectly.

    You can call the concepts of modern oncology by whatever buzzword you wish. You can describe the orbital motions of planets in our solar system with the earth as a centre, but it ends up making everything unnecessarily complex

    I agree that buzzwords and aesthetics have nothing to do with science. What I am claiming is just the opposite: that somatic evolution has great explanatory and predictive power, better than any theory that does not contain it. Therefore it should be accepted, until proven otherwise or supplanted by a still better theory.

  • Pingback: The Conflict Between Complex Systems and Reductionism « Complex Adaptive Systems()

  • Pingback: Incidentalomas « Complex Adaptive Systems()