When normal is the favoured type


Before we discuss cancer, let’s consider other sets of nasty characters – criminals, Nazis and terrorists. We know that the best way to limit crime is to create neighbourhoods that provide opportunities for education and jobs – a healthy infrastructure disfavours the criminal phenotype.  Similarly, the Marshall Plan contributed to the rebuilding of Europe after WWII, likely preventing the emergence of another despotic regime (as the Nazis did in Germany after WWI, capitalizing on a country devastated after the war).  The U.S.-led invasion of Iraq in 2002 had the stated goal of crushing terrorism, but instead left the country in such a devastated state that terrorism became a much bigger problem there than it was before the invasion.  A devastated landscape is conducive to the terrorist phenotype. In each case, the numbers of potential criminals, Nazis, and terrorists born would have been similar, independent of the prevailing environment – the distribution of human genotypes is not the key variable here. Environmental disruption and instability favour phenotypes that we would consider as “maladaptive” in a healthy environment. In contrast, a stable healthy society is much more conducive to normality.

Early work from Beatriz Mintz, Mina Bissell and others has shown how the tissue microenvironment can substantially influence the cancer phenotype – the same cancer cells were shown to contribute to functional tissue phenotypes in a normal context, but form an invasive cancer in the wrong environment (1,2). In these cases, the environment directly alters the cellular phenotype. For the last 15 years or so, my lab has focused more on how environments either select for, or importantly, select against oncogenic phenotypes.  If we want to understand why we get cancers, we should start by understanding why cancers are so rare during youth.  We have argued that young, healthy tissues are inherently tumour suppressive, by disfavouring selection for oncogenic phenotypes (not so subtly presaged in the preceding paragraph).

You might ask – “how can this be? Oncogenic mutations will cause proliferation, even in a young healthy tissue. Don’t we avoid cancer in youth simply because insufficient mutations have accumulated?” While these views are widespread, they are wrong. I’ve dealt with the second belief extensively in previous critiques of this view (3,4), so we’ll focus on the first one here. That the occurrence of an oncogenic mutation should provide a cell-autonomous advantage is largely based on decades of research in “out of context” models, such as in cell cultures. Moreover, many mouse models engineer the oncogenic mutation in all of the cells of the targeted tissue (bypassing any need for the mutation to be advantageous).  What happens when oncogenic mutations are created in normal stem cells in healthy tissues?  For mouse hematopoietic stem cells (HSC), the result is quite reproducible across mutations disrupting tumour suppressor genes or activating oncogenes – these mutations cause loss of self-renewal (5). Loss of self-renewal is the kiss of death for a stem cell clone, leading to clonal exhaustion through differentiation (and the faster the rate of division, the faster the exhaustion).

We have argued that the same evolutionary force – stabilizing selection – that limits changes in organisms when environments are stable, also functions to limit somatic evolution in our tissues. In 1948, Herman J. Muller described how fruit flies maintain very constant features, such as the morphology and position of their wings (6). While it was very easy to identify fly mutants that changed these traits, in each case, mutational changes appeared to reduce fly fitness. He described the “high adaptive value of precisely the ‘normal’ degree of gene expression now existing” and that phenotypic stability in flies across many generations is “due to active selection in favour of the normal type”.

Just like a healthy forest will disfavour evolutionary change, so does a healthy tissue favour the status quo. Evolution is not progressive, and it does not seek any particular goal. Stabilizing selection is often depicted as a bell shaped curve, whereby decreasing or increasing a trait leads to reductions in fitness (Figure A). Perhaps the best demonstration for how such a relationship exists for HSC comes from the studies of Sean Morrison and colleagues on protein synthesis rates in HSC, showing that heterozygous mutations in a ribosomal protein (Rpl24Bst/+) reduce translation in HSC and impair HSC function and competitive potential (7). Morrison and others had previously shown that deletion of the PTEN tumour suppressor gene reduces HSC self-renewal (8,9); while the mutation was engineered in all HSC, were this mutation to occur spontaneously in a single HSC, it should almost always lead to clonal exhaustion (at least in a healthy HSC pool). PTEN loss also increases protein synthesis rates in HSC. Strikingly, combining PTEN loss with ribosomal protein mutation heterozygosity restores more normal translation in HSC, and also restores HSC function (7) (Figure B). Thus, both increases and decreases in protein synthesis rates impairs the ability of an HSC to be maintained in the compartment. Very similar relationships are evident for reactive oxygen species (ROS) (10) and Myc (11), where too much or too little of either leads to reduced HSC activity. Increasing translation, creating more ROS, and activating Myc – these are all common features of many oncogenic mutations, and they each have been shown to impair HSC self-renewal and/or competitive potential.


While some oncogenic mutations induced in HSC have been shown to increase competitiveness and self-renewal, I would argue that the methods used need to be considered. In particular, studies demonstrating that loss of DNMT3A and TET2 tumour suppressor genes increase HSC competitiveness have used high dose irradiation to condition recipient mice for these assays (12-18). I propose that these mutations will be either much less advantageous, or even disadvantageous, in healthy unperturbed bone marrow. Many other studies have utilized Mx-CRE to activate oncogenic mutations, which comes with the caveat that recombination requires induction of a potent interferon response (usually via injection of poly-IC). Ideally, we need to induce rare oncogenic mutations in healthy tissues without system disruption if we really want to understand how these mutations impact cell fate (not easy to do).

There are other ways to disfavour deviation from “normal”, including through induction of apoptosis or senescence following damage or aberrant activation of oncogenic signalling (19). Starting in the 1970’s, Gines Morata and colleagues described how cell competition in the fruit fly Drosophila melanogaster can eliminate cells with reduced function, such as through reduced protein synthesis or lower levels of Myc (20,21). Less competitive cells are actively eliminated (and even eaten) by their more fit neighbours. More recent studies have shown that too much of a normally good activity can be a bad thing too. Cells with oncogenic mutations in fly cells are actively eliminated by their wild-type neighbours (22,23), through well conserved signalling pathways. Again, the “normal type” is favoured. Similar mechanisms are evident in mammals, whereby an oncogenically mutated cell is forced out of a cell monolayer by its wild-type neighbours (24). For a cell in our airways or in our guts, such a fate would lead to expulsion of the covenant-disobeying cell into the outside world. A recent study further described how a normal tissue recognizes aberrant growths (whether due to impaired function or hyperfunction), with normal cells enveloping the mutant cells and expelling them from the tissue (25). Tumour suppression by eviction!

So basically, we are proposing that millions of years of evolution have selected for stem cells that are well-adapted to their tissue niches, which will strongly disfavour mutations that change phenotype. So why then do we ever get cancer? Importantly, the maintenance of healthy tissues, which favours the normal phenotype of resident stem cells, has been strongly selected by evolution through periods of likely reproductive success. However, as the odds of successful survival and reproduction (including the rearing of offspring for animals like humans) declines due to predation, low food availability, weather or other causes, the selective pressure to maintain tissues declines. So the same ideas proposed over 50 years ago to explain physiological decline in old age (26,27) can be applied to understand why cancer incidence also increases dramatically in older ages (28). The same oncogenic mutations that may be maladaptive in a healthy youthful tissue can become adaptive in an aged tissue, by providing an adaptation to the age-altered tissue landscape. Modern exposures, such as from cigarette smoke, pollution or alcohol, or more recent epidemics such as obesity, also clearly change tissue landscapes, and thus will radically change selective pressures. Hence, while it has been estimated that cancer incidence for most animals in the absence of recent disruption will typically be less than 5% (29), about 40% of humans will experience cancer, due both to longer lives and to modern exposures and lifestyles. In fact, just about everything we know that decreases cancer risk (exercise, a good diet, not smoking, etc.) is associated with healthier tissues – better favouring the normal phenotype for resident stem and progenitor cells.

While national security based on bombing other countries and “tough on crime policies” can earn you votes at home, building schools and infrastructure will produce better and less costly results in the long run. We similarly need to develop approaches to modify tissue landscapes such that the malignant phenotype is disfavoured, whether for prevention or treatment. Understanding that the fitness impact of a malignant phenotype is highly dependent on microenvironmental context should shift the emphasis towards targeting tissue landscapes, complementing therapies that are directly targeting malignant cells.



1. Mintz B, Illmensee K. Normal genetically mosaic mice produced from malignant teratocarcinoma cells. Proc Natl Acad Sci USA 1975;72:3585-9

2. Bissell MJ, Hines WC. Why don’t we get more cancer? A proposed role of the microenvironment in restraining cancer progression. Nature medicine 2011;17:320-9

3. DeGregori J. Connecting cancer to its causes requires incorporation of effects on tissue microenvironments. Cancer research 2017

4. Liggett LA, DeGregori J. Changing mutational and adaptive landscapes and the genesis of cancer. Biochimica et biophysica acta 2017

5. DeGregori J. Challenging the axiom: does the occurrence of oncogenic mutations truly limit cancer development with age? Oncogene 2012;32:1869-75

6. Muller HJ. Evidence of the precision of genetic adaptation. Harvey Lecture Series 1948;XLIII:165–229

7. Signer RAJ, Magee JA, Salic A, Morrison SJ. Haematopoietic stem cells require a highly regulated protein synthesis rate. Nature 2014;509:49-54

8. Yilmaz OH, Valdez R, Theisen BK, Guo W, Ferguson DO, Wu H, et al. Pten dependence distinguishes haematopoietic stem cells from leukaemia-initiating cells. Nature 2006;441:475-82

9. Zhang J, Grindley JC, Yin T, Jayasinghe S, He XC, Ross JT, et al. PTEN maintains haematopoietic stem cells and acts in lineage choice and leukaemia prevention. Nature 2006;441:518-22

10. Ludin A, Gur-Cohen S, Golan K, Kaufmann KB, Itkin T, Medaglia C, et al. Reactive oxygen species regulate hematopoietic stem cell self-renewal, migration and development, as well as their bone marrow microenvironment. Antioxidants & redox signaling 2014;21:1605-19

11. Wilson A, Murphy MJ, Oskarsson T, Kaloulis K, Bettess MD, Oser GM, et al. c-Myc controls the balance between hematopoietic stem cell self-renewal and differentiation. Genes & development 2004;18:2747-63

12. Cimmino L, Dolgalev I, Wang Y, Yoshimi A, Martin GH, Wang J, et al. Restoration of TET2 Function Blocks Aberrant Self-Renewal and Leukemia Progression. Cell 2017

13. Kunimoto H, Fukuchi Y, Sakurai M, Takubo K, Okamoto S, Nakajima H. Tet2-mutated myeloid progenitors possess aberrant in vitro self-renewal capacity. Blood 2014;123:2897-9

14. Moran-Crusio K, Reavie L, Shih A, Abdel-Wahab O, Ndiaye-Lobry D, Lobry C, et al. Tet2 loss leads to increased hematopoietic stem cell self-renewal and myeloid transformation. Cancer cell 2011;20:11-24

15. Challen GA, Sun D, Jeong M, Luo M, Jelinek J, Berg JS, et al. Dnmt3a is essential for hematopoietic stem cell differentiation. Nature genetics 2012;44:23-31

16. Haladyna JN, Yamauchi T, Neff T, Bernt KM. Epigenetic modifiers in normal and malignant hematopoiesis. Epigenomics 2015;7:301-20

17. Ko M, Bandukwala HS, An J, Lamperti ED, Thompson EC, Hastie R, et al. Ten-Eleven-Translocation 2 (TET2) negatively regulates homeostasis and differentiation of hematopoietic stem cells in mice. Proceedings of the National Academy of Sciences of the United States of America 2011;108:14566-71

18. Quivoron C, Couronne L, Della Valle V, Lopez CK, Plo I, Wagner-Ballon O, et al. TET2 inactivation results in pleiotropic hematopoietic abnormalities in mouse and is a recurrent event during human lymphomagenesis. Cancer cell 2011;20:25-38

19. Lowe SW, Cepero E, Evan G. Intrinsic tumour suppression. Nature 2004;432:307-15

20. Morata G, Ripoll P. Minutes: Mutants of Drosophila autonomously affecting cell division rate. Developmental biology 1975;42:211-21

21. Baker NE. Cell competition. Curr Biol 2011;21:R11-5

22. Ballesteros-Arias L, Saavedra V, Morata G. Cell competition may function either as tumour-suppressing or as tumour-stimulating factor in Drosophila. Oncogene 2014;33:4377-84

23. Menendez J, Perez-Garijo A, Calleja M, Morata G. A tumor-suppressing mechanism in Drosophila involving cell competition and the Hippo pathway. Proceedings of the National Academy of Sciences of the United States of America 2010;107:14651-6

24. Kajita M, Fujita Y. EDAC: Epithelial defence against cancer-cell competition between normal and transformed epithelial cells in mammals. Journal of biochemistry 2015;158:15-23

25. Brown S, Pineda CM, Xin T, Boucher J, Suozzi KC, Park S, et al. Correction of aberrant growth preserves tissue homeostasis. Nature 2017;548:334-7

26. Medawar P. An unsolved problem of biology. London: H.K.Lewis; 1952.

27. Williams GC. Pleiotropy, Natural Selection, and the Evolution of Senescence. Evolution 1957;11:398-411

28. Rozhok AI, DeGregori J. The evolution of lifespan and age-dependent cancer risk. Trends Cancer 2016;2:552-60

29. Hochberg ME, Noble RJ. A framework for how environment contributes to cancer risk. Ecology Letters 2017:20:117-134


How cancer outsmarts multicellularity

David Goode & Anna Trigos

Peter MacCallum Cancer Centre, Melbourne, Australia

Title image


The transition from unicellularity to multicellularity was one of the most significant advances in the evolution of life on Earth. Enhanced cooperation and communication between cells enabled the development of increasingly complex and specialized tissues; the resulting differentiation and division of labour enabled the acquisition of a broad array of new features. In turn, this allowed adaptation to a wide range of new ecological niches, rapidly accelerating the pace of evolution and creating an explosion of diversity across the plant, animal and fungal kingdoms.

This array of new multicellular phenotypes was driven by major innovations on the molecular level. Coordinating the growth of millions or billions of cells required the evolution of new proteins and regulatory elements to impose control over core cellular processes such as translation, DNA replication and the cell cycle. The activity of these primitive processes could then be temporally and spatially ordered, rather than stochastically responding as individual cells to external environmental cues.

In this sense, multicellularity can be viewed as a result of a molecular evolutionary process of selection for cooperative rather than competitive growth, via a default switching off of proliferative pathways in the majority of cells. However, it was also necessary to maintain the means to activate such pathways under certain conditions, e.g., embryogenesis, wound healing, or immune response. This requirement meant newly-evolved metazoan regulatory mechanisms had to be bidirectional. Such flexibility was advantageous, but also provided a ‘back-door’ that allows cells to inappropriately reactivate primitive processes and thus to enable proliferation independently of normal control structures.

Unchecked, this leads to a scenario where selection for the survival of individual cells is favoured over survival of the entire organism, namely cancer. A more formal framework for this reversion is the Atavism hypothesis of cancer1,2, which proposes that cancer cells lose their identity as cells of a multicellular tissue, and start to behave more like unicellular organisms. This is achieved under strong selection for reactivation of gene expression programmes that date back to unicellular ancestors, manifested as a loss of communication with neighbouring cells, and loss of differentiation and tissue structure.

The atavism hypothesis of cancer was largely founded on the broad phenotypic similarities between cancer cells and unicellular species such as bacteria and yeast. More recently, the molecular changes driving these phenomena have begun to come to light. Among them are the finding that many of the genes commonly involved in cancer date back to the emergence of the first metazoans3 and increasingly malignant tumour phenotypes are accompanied by progressive mutation of such genes4.

Recently, we completed a comprehensive analysis of potential atavistic signatures in the transcriptomes of 7 solid tumour types using data from The Cancer Genome Atlas5. We observed a strong and consistent increase in the expression of genes of unicellular organisms across all tumour types, with a concomitant decrease in expression of more recently acquired genes5; an observation that is consistent with increased importance in cancer of the most highly conserved genes in the genome. These results imply a compartmentalisation of gene expression by evolutionary age in tumour cells, with a significant separation of the activity of genes of unicellular and multicellular ancestry. This effect is quite clear for genes involved in ancient processes throughout the tree of life, such as protein translation and cell cycle progression, as expected. But even the  expression of genes of unicellular origin that have been co-opted into processes associated with multicellularity, such as cell-cell adhesion and organogenesis, was maintained or even increased in tumours.

Together these observations suggest an increased reliance on the more primitive parts of the transcriptome during tumourigenesis. How might this occur? We overlaid interaction data onto TCGA expression data, revealing several pairs of highly connected unicellular and multicellular processes whose expression went from highly positively correlated in normal cells to highly negatively correlated in tumours. This indicated a loss of the regulatory mechanisms coordinating the expression of certain processes in tumours, leading to mutual exclusivity between those processes, presumably to the advantage of cancer. The genes apparently mediating these switches were both unicellular and multicellular, but overall enriched for genes essential for cancer cell growth, based on functional genomics screen data from Project Achilles6.

We interpret this mutual exclusivity as a consequence of how metazoan gene regulatory networks were laid out during evolution, and how they get disrupted and rewired in cancer. The core of the network was formed during the emergence of the earliest unicellular organisms. The high degree of correlation in expression among genes involved in unicellular processes indicates these components are highly connected, and robust to perturbation. The evolution of multicellularity built an outer layer around the conserved ancient core, with key genes linking the two and providing regulatory control. When these links are broken, there is uncoupling of the unicellular and multicellular halves of the network, leading to a more proliferative, more ‘primitive’ phenotype, and malignant growth. However, cancer cells are not simply hijacking the wiring generally used by these other normal cell types when losing their cell complexity, as the specific processes contributing to the atavistic state are markedly different.

The development of therapeutic strategies derived from an evolutionary perspective opens the possibility of a streamlined approach to the discovery of novel gene targets and novel drug repurposing strategies7. However, to achieve this we need to further refine our understanding of the mechanisms driving the uncoupling of unicellularity and multicellularity in the development and progression of cancer, such as how genetic and epigenetic alterations modulate a loss of multicellularity. Furthermore, by expanding our understanding of the association between robustness to perturbation and evolutionary history we would be better poised to predict genes or pathways of resistance, and develop treatment strategies that a priori incorporate this knowledge.

Given their essentiality for cell viability, the fundamental cellular processes common to both unicellular and multicellular life are characterised by plasticity in gene interactions and the presence of redundant pathways. Thus ensuring that, even under severe insults (e.g. stress or drug treatment), cell viability remains. This enhanced plasticity and robustness in the core of the network evolved early on, explaining why many drug resistance mechanisms involve unicellular genes and processes. On the other hand, the more recent genes related to multicellular-specific processes may only be required in certain situations, and therefore, there is less selective pressure to maintain their integrity. From a therapeutic perspective, identifying and targeting gene network regions that play key roles in tumour development but generally have lower resilience to insults would be more likely to delay the onset of drug resistance.

The genes at the interface between unicellular and multicellular regulatory networks represent promising therapeutic targets, as they could signal sites of vulnerabilities that, when targeted, would cause widespread disruption in molecular networks and hence lead to cell death7. Specificity to cancer cells could be obtained by focusing on genes mediating mutually exclusive associations between unicellular and multicellular processes unique to cancer, such as those we recently identified5. There may also be potential to repurpose existing drugs to exploit the mutual exclusivity between unicellular and multicellular processes. By this means, drugs promoting the activation of multicellular processes would reduce the activity of unicellular processes, with the aim of a reduction in malignancy due to reactivation of multicellular features. A more pragmatic approach in the near term may be to attack weaknesses in cancer brought about by the loss of particular multicellular genes. This has been proven effective in tumour cell lines deficient in cysteine/glutamate antitransporter activity, as a means to manipulate intracellular oxidative stress and affect cell survival8.

Hundreds of millions of years of evolution have guided the formation of robust and flexible genetic and protein interaction networks in modern metazoan cells to maintain the diverse sets of functionalities required for multicellular life. Selection for maintenance of multicellular control structures is favoured over the long term, but in the short term, the drive of individual cells to multiply and spread can win out, with disastrous consequences for some unfortunate individuals. Understanding the forces that have shaped the molecular basis of multicellularity and the countervailing forces that can undo them will offer crucial insights into the fundamental nature of cancer and the potential for smarter, more efficient treatment options for patients.



  1. Davies PC, Lineweaver CH: Cancer tumors as Metazoa 1.0: tapping genes of ancient ancestors. Phys Biol 2011, 8:015001.
  2. Vincent M: Cancer: a de-repression of a default survival program common to all cells?: a life-history perspective on the nature of cancer. Bioessays 2012, 34:72-82.
  3. Domazet-Loso T, Tautz D: Phylostratigraphic tracking of cancer genes suggests a link to the emergence of multicellularity in metazoa. BMC Biol 2010, 8:66.
  4. Chen H, Lin F, Xing K, He X: The reverse evolution from multicellularity to unicellularity during carcinogenesis. Nat Commun 2015, 6:6367.
  5. Trigos AS, Pearson RB, Papenfuss AT, Goode DL: Altered interactions between unicellular and multicellular genes drive hallmarks of transformation in a diverse range of solid tumors. Proc Natl Acad Sci U S A 2017, 114:6406-6411.
  6. Aguirre AJ, Meyers RM, Weir BA, Vazquez F, Zhang CZ, Ben-David U, Cook A, Ha G, Harrington WF, Doshi MB, et al: Genomic Copy Number Dictates a Gene-Independent Cell Response to CRISPR/Cas9 Targeting. Cancer Discov 2016, 6:914-929.
  7. Trigos AS, Pearson RB, Papenfuss AT, Goode DL: How the evolution of multicellularity set the stage for cancer. Br J Cancer 2018 [Article in Press].
  8. Liu DS, Duong CP, Haupt S, Montgomery KG, House CM, Azar WJ, Pearson HB, Fisher OM, Read M, Guerra GR, et al: Inhibiting the system xC_/glutathione axis targets cancers with mutant-p53 accumulation. Nat Commun 2017, 8:14844.

 Featured image: http://firstwestcapital.ca/wp-content/uploads/iStock-153724177.jpg

Which Cancers are Most Survivable and Why?

Cancer is not a death sentence; there is a great deal of difference between a prostate and a pancreatic cancer diagnosis, and even differences between subtypes of cancer within any particular organ. Recent statistics on cancer survival rates are instructive (Table 1).


Table 1. The most survivable cancers according to the US SEER database of cancers diagnosed between 2005 and 20111.

Cancer Type Median age at diagnosis 5-year relative survival
Skin (basal & squamous) unknown 99.9%
Prostate 66 99%
Thyroid 50 98%
Testis 33 95%
Melanoma of the skin 63 92%
Breast (female) 61 89%
Hodgkin Lymphoma 38 86%
All childhood cancers 0-14 83%
All cancers (excluding skin) 65 67%


How can we understand this?


When, in the natural course of a cancer, do you feel sick?

If the symptoms of cancer appear early in the natural course of the disease, it is generally curable. There are two reasons for this. First, if you detect a cancer before it has escaped the reach of a surgeon’s knife, it can be removed. End of story.

Second, even if it can’t be cut out as is the case with most blood cancers, if the cancer has only had a few years to accumulate mutations, it is less likely to have acquired mutations that will make it resistant to a therapy. In contrast, if you don’t feel sick until very late in the process, as in pancreatic cancer, it is likely to have acquired resistance mutations and no matter what you try, you are unlikely to cure the disease. The result, in pancreatic cancer, is a 7% 5-year relative survival rate1.

There is an evolutionary theory for why solid cancers are generally more lethal than blood cancers of the immune system; there appears to be more checks and tumour suppressor mechanisms in place to prevent solid cancers than there are to prevent immune cell cancers. With immune cell cancers, typically only a few genes need to mutate in order for you to feel sick. Therefore, when a patient shows up with a leukemia or lymphoma, there are fewer mutations in their cells, and so there is less chance that a mutant cell has already acquired resistance to the coming therapy. In contrast, everything but the kitchen sink has to break in order for cells in most organs to grow out of control. With so many different things broken, and such a diversity of mutant cells, it is no wonder that resistant cells often lurk somewhere in a solid cancer.

There is an important implication of this insight. We should be developing measurements of the diversity within cancers to help guide the management of those cancers. For tumours with low diversity, we have a better chance of achieving a cure through therapy2. But for tumors with lots of diversity, we need to consider how to manage the therapeutic resistance that is most likely already present, perhaps through strategies like adaptive therapy3,4. Of course, if a tumour has not yet metastasised, surgery can effectively avoid the whole issue of the evolution of therapeutic resistance.


Surgeons cure more cancer than oncologists

Or, to put it another way, oncologists have a more difficult problem than surgeons. Their drugs must find and kill every last cancer cell, no matter where it is hidden and what mutations it carries.

Skin cancers are extremely common but easy to surgically remove. Basal and squamous cell carcinomas of the skin are so common, they are often excluded from studies and cancer registries, like the SEER database that was used to produce most of Table 1. Approximately 5.4 million skin cancers (other than melanoma) are diagnosed in the US each year5, making up about 75% of all newly diagnosed cancers. However, only about 2,000 Americans die from them each year6 because the vast majority are detected before they metastasise and can be removed.

The surprise for me in Table 1 was melanoma. Melanoma is infamous for being one of the most mutated of all cancers (along with lung cancer)7,8. Those mutations are the legacy of UV light and smoking. In addition, melanoma readily metastasises, and unlike most cancers, can spread anywhere in the body. Yet, it is listed as the fourth most survivable cancer. Because melanomas are exposed on the skin, we have the opportunity to see them every day in the mirror, and catch them early. The result is that 84% of melanomas are diagnosed before they metastasise and can be cured surgically, with a 98% 5-year relative survival rate.


The benefits of indolence and hormones

Some cancers are so slow growing that we can live with them without them killing us; they are indolent. This is famously true for both prostate cancer and thyroid cancer9. In the US, autopsy studies have revealed that 80% of men over the age of 70 have some cancer hanging out in their prostates, but few of them will die from this10. Small nodules of cancer in the thyroid are so common they are considered “normal”11. Autopsy studies have found minute nodules of thyroid cancer in 8% of the general population12, however, it rarely discovers a way to generate blood vessels to feed itself, and so it never grows large enough to harm us.

There are extensive screening programmes for both breast and prostate cancer. Unfortunately, there is an inherent bias in screening programs. They preferentially find the slowest growing tumours because those are the ones hanging around for years, available to be detected. In contrast, the fast-growing tumours can pop up and make us sick before we ever have a chance to detect them through a regular screen. This implies that many of the cancers we detect, many of the cancers in Table 1, would never kill us even if they were never treated. The survival statistics in Table 1 are inflated by indolence.

Hormones are also part of the story. The cells in most prostate and breast cancers need hormones (testosterone and estrogen) in order to reproduce. When we deny them those hormones, they stop growing, and in many cases they start behaving like they are starving, slowly consuming themselves. It takes a long time, if ever, for some of those cells to discover ways to live without those hormones, and so survival times are much longer for breast and prostate cancers than for other solid cancers. This does suggest that if we are able to deny growth factors to other cancers, if we could stop their growth, rather than trying to kill them, we might be able to increase survival times in those cancers as well.

There is another piece to the puzzle of thyroid cancer. While it is generally detected when it is small enough to be removed surgically, it also has a particular Achilles heel. Because thyroid tissue is the only tissue that uses iodine, treatment with radioactive iodine efficiently targets any remaining thyroid tissue (and cancer) after surgery.


To be young and unmutated

In addition to being slow growing and dependent on iodine, thyroid cancer typically carries few mutations, probably because it is generally detected at relatively young ages. So, resistance mutations are less likely to be present at diagnosis, compared to highly mutated cancers.

Testicular cancer, Hodgkin lymphoma and the childhood cancers are all detected at young ages. In general, like thyroid cancer, this is associated with the accumulation of fewer mutations8,13, and little genetic diversity. Since the success of systemic therapies, such as chemotherapy or targeted agents, depends on the absence of resistance mutations, these genetically homogeneous cancers are more likely to be curable than genetically diverse cancers2,14


The artefacts of technology

Don Pinkel, pioneer of childhood leukaemia treatment, has long argued that what makes for a good or bad cancer is mostly an artefact of treatment. Can we detect it when it has few mutations? Are we detecting things that aren’t really lethal cancers? Do we currently have good treatments for it? This all changes with advances in medicine and technology. A sizable minority of late stage melanomas and lung cancers can now be cured by immune blockade therapy, particularly the highly mutated tumors. This goes against much of what I’ve said about the evolution of therapeutic resistance, but for a good reason. Highly mutated cancers produce more abnormal proteins that the immune system can recognize as non-self. Thus, the lineup of the most survivable cancers will change in the future. What won’t change, is the need to deal with their evolution.



  1. Miller, K. D. et al. Cancer treatment and survivorship statistics, 2016. CA Cancer J Clin 66, 271-289, doi:10.3322/caac.21349 (2016).
  2. Morris, L. G. et al. Pan-cancer analysis of intratumor heterogeneity as a prognostic determinant of survival. Oncotarget 7, 10051-10063, doi:10.18632/oncotarget.7067 (2016).
  3. Gatenby, R. A., Silva, A. S., Gillies, R. J. & Frieden, B. R. Adaptive therapy. Cancer Res 69, 4894-4903, doi:69/11/4894 [pii] 10.1158/0008-5472.CAN-08-3658 (2009).
  4. Enriquez-Navas, P. M. et al. Exploiting evolutionary principles to prolong tumor control in preclinical models of breast cancer. Sci Transl Med 8, 327ra324, doi:10.1126/scitranslmed.aad7842 (2016).
  5. Rogers, H. W., Weinstock, M. A., Feldman, S. R. & Coldiron, B. M. Incidence Estimate of Nonmelanoma Skin Cancer (Keratinocyte Carcinomas) in the U.S. Population, 2012. JAMA Dermatol 151, 1081-1086, doi:10.1001/jamadermatol.2015.1187 (2015).
  6. Weinstock, M. A. et al. Nonmelanoma skin cancer mortality. A population-based study. Arch Dermatol 127, 1194-1197 (1991).
  7. Alexandrov, L. B. et al. Signatures of mutational processes in human cancer. Nature 500, 415-421, doi:10.1038/nature12477 (2013).
  8. Lawrence, M. S. et al. Mutational heterogeneity in cancer and the search for new cancer-associated genes. Nature 499, 214-218, doi:10.1038/nature12213 (2013).
  9. Bissell, M. J. & Hines, W. C. Why don’t we get more cancer? A proposed role of the microenvironment in restraining cancer progression. Nat Med 17, 320-329 (2011).
  10. Haas, G. P., Delongchamps, N., Brawley, O. W., Wang, C. Y. & de la Roza, G. The worldwide epidemiology of prostate cancer: perspectives from autopsy studies. Can J Urol 15, 3866-3871 (2008).
  11. Harach, H. R., Franssila, K. O. & Wasenius, V. M. Occult papillary carcinoma of the thyroid. A “normal” finding in Finland. A systematic autopsy study. Cancer 56, 531-538 (1985).
  12. Valle, L. A. & Kloos, R. T. The prevalence of occult medullary thyroid carcinoma at autopsy. J Clin Endocrinol Metab 96, E109-113, doi:10.1210/jc.2010-0959 (2011).
  13. Litchfield, K. et al. Whole-exome sequencing reveals the mutational spectrum of testicular germ cell tumours. Nat Commun 6, 5973, doi:10.1038/ncomms6973 (2015).
  14. Bochtler T, Stölzel F, Heilig CE, Kunz C, Mohr B, Jauch A, Janssen JWG, Kramer M, Benner A, Bornhäuser M, Ho AD, Ehninger G, Schaich M, Krämer A (2013)  Clonal heterogeneity as detected by metaphase karyotyping is an indicator of poor prognosis in acute myeloid leukemia.  J Clin Oncol, 31: 3898-3905.

Do mutations cause cancer? (or the dog that did not bark)

Evolution is change over time, and it is well-accepted that cancers evolve through the stepwise accumulation of somatic mutations. Logically, mutations ‘cause’ cancer, and therefore, simplistically, the key to preventing cancer could be to avoid mutations. However, epithelium, like the skin and intestines, divide and shed millions of cells every day, and could accumulate many mutations because DNA replication is imperfect.

One potential safeguard against ‘replication’ errors is a stem cell hierarchy, where long-lived stem cells divide infrequently. However, studies in mice indicate that both skin1 and intestinal stem cells2 are not quiescent but rather are actively dividing. Such tissues are primed for evolution because many more cells are produced than can survive.

Continue reading

Lessons from cancer risk in animals


Think for a moment about a young girl, four years old, diagnosed with acute lymphoblastic leukaemia (ALL). This is not so unusual, and in fact, ALL is the most commonly diagnosed childhood cancer. However, only two years earlier, this young girl was also diagnosed with a grade II glioma in her brain, treated by surgical resection with no chemotherapy or radiation exposure. Moreover, her father and her father’s brother both recently died of aggressive glioblastoma multiforme (GBM) brain tumours. This young girl and her family have Li-Fraumeni Syndrome, an inherited defect in one of the TP53 genes leading to a nearly 100% lifetime risk of cancer1.

Continue reading

Ways of escape

Cancers are life threatening because they migrate within the body, spreading far from their point of origin. This process – metastasis – hijacks tissues and compromises their critical functions. When they reach this stage, most cancer clones will be robust and resistant to treatment, whether that be radiotherapy, chemotherapy or immunotherapy. So, in a sense, it is resistance that is the major stumbling block to successful treatment. Those exceptional cancers that are curable, even when disseminated (childhood acute lymphoblastic leukaemia, testicular cancer and choriocarcinoma) retain sensitivity.

Continue reading

“Big Bang” cancer growth

In the second of our guest blog posts, Dr Andrea Sottoriva describes how a comparison between the expanding universe and the growth of cancers led him to formulate his “Big Bang” theory of tumour growth – a model with novel treatment implications.

Image from Physicsworld.com

Image from Physicsworld.com

In 1929 Edward Hubble, sitting at the top of Mount Wilson, observed that stars and galaxies are moving away from each other. He reasoned that, if stars are continuously moving apart, they must have been closer together at earlier times, to the point that at the very beginning the entire cosmos would have been compressed into a tiny space. This led to the hypothesis that our universe could have originated from a cosmic explosion, “the Big Bang”. But where are the remnants of such an enormous blast? Surely such a phenomenon must have left its mark in today’s universe? In fact, it did. Radio astronomers, Arno Penzias and Robert Wilson, detected the Cosmic Microwave Background radiation in 1964. This is the glow of the Big Bang explosion, it permeates the whole universe at an almost uniform -270 degrees Celsius.

So, what does all of this have to do with cancer? Tumours are large collections of cancer cells that grow out of control and invade healthy tissue, thus becoming life-threatening. Like the universe, cancers expand from something tiny, a single tiny cell. By sequencing the DNA of tumours we discovered that each cancer is unique to a single patient, in the same way that the universe is unique, as far as we can tell.

Continue reading

Convergence in tumour evolution: singing the same tune

© Katerina Kousalova/ Dreamstime.com/ License: Royalty Free

© Katerina Kousalova/ Dreamstime.com/ License: Royalty Free

Cancer clone evolution, just like evolutionary speciation, is characterised by an extraordinary diversity of descendants derived from a common ancestor. Yet, paradoxically, some evolutionary trajectories are convergent on a common phenotype.

The classical examples of convergency from evolutionary biology include eyes, wings, big brains and social structures, all of which have been ‘invented’ multiple times, independently. We find that their genetic, developmental and biochemical details are often distinct but in the end, the functional result is very similar 1.

Continue reading

Cancer versus immunological diversity

©Jakedan/ Dreamstime.com/ License: Royalty Free

©Jakedan/ Dreamstime.com/ License: Royalty Free

We are seeing a renaissance of optimism about immunotherapy for cancer – after many years of disappointment. Patients with advanced and clinically intransigent lung cancers and melanomas, treated in early clinical trials with antibodies to immune checkpoint inhibitors PD-1 and CTLA-4, have been surviving longer than would previously have been expected 1,2. And other studies have demonstrated that patients whose tumours were infiltrated with lymphocytes show better outcomes 3.

Putting these observations together, the inference is that some tumours present neoantigens that are recognised by the immune system and that this reactivity can be boosted by releasing the checkpoint brakes on the immune system.

Continue reading

Cancer Evolution – It’s in the Blood!

In the first of our guest blog posts, Dr. Marco Gerlinger highlights some of the remarkable developments being made in ctDNA analysis, a powerful new technology with the potential to transform tumour predictions and treatment outcomes.

Photo credit: Csutkaa via Foter.com / CC BY-NC-SA

Photo credit: Csutkaa via Foter.com / CC BY-NC-SA

Cancer cells are masters in adapting to changing environments. This allows them to colonise other organs, to form metastases and also to acquire drug resistance. Darwinian evolution is thought to be a key driver of this adaptability. Randomly acquired mutations encode for novel phenotypes and some of these phenotypes may allow individual cells to survive changes in the environment1.

This adaptability is a key reason for the high rates of mortality from metastatic cancers. Treating a cancer that cannot evolve would probably be an easy task – maybe as straightforward as eradicating a bacterial infection with antibiotics. Thus, there is great need to understand how and why cancers readily evolve and to use this information to design more effective treatment approaches for ever-changing cancers.
Continue reading