Cancer as a Complex Adaptive System

Heng, et al recently published a review paper that brings together and touches on many different aspects of cancer complexity.  I thought this an opportunity to selectively quote the paper and organize the quotes loosely around various complex systems concepts they relate to.  I’m curious whether this makes sense to readers of this blog, or whether there’s too much unexplained jargon and too many large conceptual leaps.  Please ask questions or make comments freely below.

One preface I think will help is to understand that genome, karyotype and chromosome refer roughly to the same thing.  Here are several schematics that I will present without explanation that together illustrate how genes relate to genome/karyotype/chromosome structure, and how that in turn relates to the so-called genetic network (loosely equivalent to the “proteome”).  Of course “gene” is an outdated and inaccurate concept, so don’t get too hung up looking for genes here, just understand that they are sub-structural elements of the genome.

From MSU website



From Heng’s paper titled”The genome-centric concept: resynthesis of evolutionary theory”

Now onto the paper.  I’ll point out that I’ve eliminated the scholarly references in the original text simply for clarity, but I don’t want readers to think that the authors have not properly credited the research that goes into the statements/claims made below.  If you’d like to read the original paper, email Henry Heng whose address is on the abstract above.  Also note that all emphasis in the quotes below is mine.

Somatic Evolution

…cancer progression is an evolutionary process where genome system replacement (rather than a common pathway) is the driving force.

It has become clear that a correct theoretical framework for cancer research is now urgently needed and the concept of somatic evolution represents just such a framework.

The increased NCCA frequencies reflect increased survival advantage while increased CCAs reflect a growth advantage. [NCCAs and CCAs are chromosomal aberrations, like gene mutations but at the genome level]

This last quote is reminiscent of the RNA autocatalysis experiments reported on earlier this year which showed divergent evolution towards two co-existing phenotypes, one that more quickly gobbled up available resources and another that was more efficient at using resources to reproduce quicker.  Perhaps there is a basic principle at work in both systems (autocatalytic RNA populations and somatic cell populations).

Instability / Heterogeneity / Diversity

Clearly, as there is no defined cancer genome (the vast majority of cancer cases display different karyotypes representing different genome systems), there is no defined cancer epigenome either.

…the most common feature in tumors is a high level of genome variation

Understanding the importance of heterogeneity is the key to understanding the general evolutionary mechanism of cancer.

…the true challenge is to understand the system behavior (stability or instability)…

When closely examining the contribution of various genetic factors, it is clear that many of the genetic loci or events are only significantly linked to tumorigenicity when they contribute to system instability (which is closely linked to genome level heterogeneity).

…it is relatively easy to establish a causative relationship between system heterogeneity and cancer evolution, as heterogeneity is the necessary pre-condition needed for cancer evolution to occur….

…instability imparts heterogeneity, which is acted on by natural selection.

The predictability of cancer can be accomplished by measuring the system heterogeneity that is shared by most patients rather than characterize each of the individual factors that contributes to cancer.

Virtual Stability / Chaotic Synchronization

Heterogeneity provides a greater chance of success that a system can adapt to the environment and survive.

…heterogeneity ‘‘noise’’ represents a key feature of bio-systems providing needed complexity and robustness.

…epigentic alteration is an initial response when the genome system is under stress.

It turns out, lower levels of ‘‘randomness’’ are essential for higher levels of regulation when facing a drastically changed environment.

In a human-centric version of a perfect world, within the multiple levels of homeostasis, environmental stress should be counteracted by epigenetic regulation; disturbances of metabolic status should be recovered; the errors of DNA replication should be repaired; altered cells should be eliminated by cell death mechanisms; abnormal clones should be constrained by the tissue architecture; and the formed cancer cells should be cleared up by the immune-system. In a cancer defined perfect world, in contrast, the break down of homeostasis is the key to success. Unfortunately, continually evolving systems are the way of life and cannot be totally prevented. In a sense, cancer is the price we pay for evolution as an interaction between system heterogeneity and homeostasis….

Facilitated Variation / Baldwin Effect

When changes are selected by the evolutionary process, these changes can be fixed either at a specific gene level or at the genome level (achieving the transition from epigenetic to genetic changes).

This is corroborated by Spencer, et al and Brock, et al, the latter of whom says, “‘pre-selection’ of non-genetic variants would markedly increase the probability of producing a random genetic mutation that may provide the basis for the survival capability of the original non-genetically variant outlier population.”

Path Dependence

…cancer cases are genetic and environmentally contingent. The pattern of specific gene mutations can only be used within a specific population with a similar genome, mutational composition as well as a similar environment.

…the stochastic events referred to here are not completely random but rather are less predictable due to differences in the initial conditions reflected by the multiple levels of genetic and epigenetic alteration.

From a system point of view, significant karyotypic changes represent a ‘‘point of no return’’ in system evolution, even though certain gene mutations and most likely epigenetic changes can influence karyotypic changes.

Upon establishment of a new genome through karyotypic evolution, it is impossible to revert back to a previous state through epigenetic alteration.

As long as the genome does not significantly change, epigenetic reprogramming could work to bring the system to its original status.

Multiple Levels

…the multiple levels of homeostasis are more important than genetic factors in constraining cancer, as alterations of system homeostasis rather than individual genetic alterations are responsible for the majority of cancers. Accordingly, the robustness of a network, the reversible features of epigenetic regulation, tissue architecture, and the immune-system will play a more important role than individual genetic alterations.

…genome level alteration within tumors is a universal feature.



Note that although physically the epigenetic level sits “above” the genome, functionally it’s really below, as indicated in this last figure.  Of course, it helps to remind ourselves that “level” is a convenient but not quite accurate concept, and they are not always clearly distinct and non-overlapping, as in this case.

Current Methodological Weaknesses

It should be noted that these weaknesses stem from an inherent paradigmatic conflict that exists in science as it’s practiced today.  These weaknesses will not be addressed until complex systems thinking pervades science in general.

…methodologies of DNA/RNA isolation and sequencing from mixed cell populations artificially average the molecular profile.

…current methods used to trace genetic loci heterogeneity are not accurate, as the admixture of DNA from different cells will wash away the true high level of heterogeneity and only display the heterogeneity of dominant clonal populations.

There is a need to change our way of thinking by focusing more on monitoring the level of heterogeneity rather than attempting to identify specific patterns in this highly dynamic process.

…the benefit of cancer intervention depends on the phase (stable or unstable) of evolution the somatic cells are in.

The strategies of attempting to reduce heterogeneity to study the mechanisms of cancer represent a flawed approach.  Without heterogeneity, there would be no cancer. That is the reason why many principles discovered using simplified homogenous experimental systems do not apply in the real world of heterogeneity.

…cancer progression is fundamentally different from developmental processes…. The terminology ‘‘cancer development’’ implies an incorrect concept and needs to be changed.

…we recommend focusing on correlation studies rather than search for a specific ‘‘causal relationship’’.

…the understanding of the overall contribution of epigenetic regulation should not focus solely on tumor suppressor genes, but rather focus on system dynamics and evolve-ability.

A true genome project would focus on the way genomic structure and topology form a genetic network and should also include epigenetic features of the genetic network.

  • Pingback: A Meditation on Biological Modeling « The Emergent Fool()

  • How does cancer kill people?

    Serious question.

  • @michael the short answer is when it interferes with enough normal bodily function, which almost always during metastasis (i.e. primary tumors rarely are the killers). Beyond that I am not qualified to explain.

    One unfortunate confounding factor is that the treatments we rely on are often pretty harmful (chemo, radiation, surgical removal), so it’s hard to say in some cases whether the patient died of the disease or the cure (or more likely some combination).

  • A researcher that I know gave me the following feedback in a private email. So as to not make waves for this person with their colleagues, I won’t identify them, but just pass on what they said, since I think it’s relevant to the discussion:

    General concern is the audience: not only does one has to be concerned about experts vs. non-experts, as expressed in your concern about jargon, but more subtly, and specifically, about addressing ‘mainstream’ cancer researcher vs. ‘outside-the-box thinkers’. Some of the new ideas you present are easily understood by the latter, but would be utterly confusing to the former who rely on a limited set of simple (simplistic) concepts carved in stone. It may be easy for you to understand the new ideas since you do not carry the baggage years of reductionist, gene-centred research, but without explicitly stating what is the current paradigm shared by the crowd and what are bold new ideas, you may confuse many readers while preaching to the choir. In other words, without providing a historical frame the arrow of innovation, the change of mind, the zeitgeist, are a bit lost.

    In terms of pedagogical background, in Fig. 1 it would help to designate what is a GENE. The concept of a gene while outdated as you correctly said, is still generally valid. The ‘outdatedness’ refers to some subtle features (splicing, regulatory RNA , etc) which are actually not very relevant. Thus it still may help to point to the DNA and indicate that a gene is a fragment of that DNA (bottom of Fig 1). Otherwise, the word ‘gene’ never appears in the background explanation. Also, in conjunction with background information, the equation “genetic network (loosely equivalent to the “proteome”) is a bit crude. I am empathetic but also know that it can easily become subject of distracting hair-splitting comments by narrow minded readers.

  • I have read several books on the mechanism of cancer, and I cannot figure it out.

    There seems to be a number of alternatives.

    1. The tumor’s physical growth interferes with local geography and signals are lost.

    2. The tumor’s demand for increased blood consumption is unstainable.

    3. The tumor is merely reflecting some failure in the nervous feedback system.

    Got any ideas what I could read to figure this out?

  • Michael, it’s such a simple and obvious question but one that gets to the heart of our collective ignorance about the disease. I ask researchers a similar question (“what is cancer?”) and the answers are not very satisfying in terms of completeness and sometimes contradictory. So it’s not surprising that your question is tough to get a satisfying answer to as well.

    Perhaps the best place that explains cancer without over-simplifying is Arny Glazier’s book.

    In my current understanding, there is an element of all three of the alternatives, and many other factors as well in terms of what kills you about cancer. In some ways it’s better to ask, how does normal multicellular life work and sustain itself? At the highest level of abstraction, I view cancer as the autocatalytic unwinding — or the photo negative — of multicellular life.

  • FoolsGold

    Often ignored are the Tumor As Fetus aspects and the Does Cancer Cause Aging or does Aging Cause Cancer question.

    I think medicine has lost sight of the fact that people who “die of caner” really die of systemic burdens imposed by pathogen proliferation.

    Oncologists focus on “reducing tumor burden” but never tell the patient that reducing the size or number of his tumors does not necessarily extend his life one bit.

  • danielhorowitz

    @FoolsGold – I think to some extent, aging causes Cancer, and to some extent Cancer causes aging. But, many things cause Cancer, and many things cause aging. So aging and Cancer are but one component of each other.

  • @rafe: My proposed definition of cancer: When cells override the body’s mechanisms that prevent them from evolving as individual entities.

    I don’t specifically know how it kills people, but such a process has obvious negative consequences.

  • @plektix, prevent whom from evolving as individual entities? And when you say “evolve” do you mean it in the technical (Darwinian) sense or do you mean develop?

  • Alex Golubev

    I’m not sure how to share this with you guys correctly, but there are a bunch of articles about chaos and complexity in the brain here:

  • I had something to say about cancer a long time ago. View on YouTube at:

  • I had something to say about cancer a long time ago. View on YouTube at:

  • M Tez

    Please read this article
    Med Hypotheses. 2006;66(4):859-60. Epub 2005 Dec 20.

    Cancer as a chaotic adaptation of the adult stem cell.
    Tez M, Koç M, Bilgin A.

    • Rafe Furst

      Why don’t you post a link to the full text?